NZ752319B2 - Using cell-free dna fragment size to determine copy number variations - Google Patents

Using cell-free dna fragment size to determine copy number variations Download PDF

Info

Publication number
NZ752319B2
NZ752319B2 NZ752319A NZ75231916A NZ752319B2 NZ 752319 B2 NZ752319 B2 NZ 752319B2 NZ 752319 A NZ752319 A NZ 752319A NZ 75231916 A NZ75231916 A NZ 75231916A NZ 752319 B2 NZ752319 B2 NZ 752319B2
Authority
NZ
New Zealand
Prior art keywords
sequence
sample
interest
chromosome
nucleic acid
Prior art date
Application number
NZ752319A
Other versions
NZ752319A (en
Inventor
Catalin Barbacioru
Gengxin Chen
Darya I Chudova
David A Comstock
Sven Duenwald
Keith W Jones
Richard P Rava
Dimitri Skvortsov
Original Assignee
Verinata Health Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US15/382,508 external-priority patent/US10095831B2/en
Application filed by Verinata Health Inc filed Critical Verinata Health Inc
Publication of NZ752319A publication Critical patent/NZ752319A/en
Publication of NZ752319B2 publication Critical patent/NZ752319B2/en

Links

Classifications

    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6869—Methods for sequencing
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q1/00—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions
    • C12Q1/68—Measuring or testing processes involving enzymes, nucleic acids or microorganisms; Compositions therefor; Processes of preparing such compositions involving nucleic acids
    • C12Q1/6876—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes
    • C12Q1/6883—Nucleic acid products used in the analysis of nucleic acids, e.g. primers or probes for diseases caused by alterations of genetic material
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2537/00—Reactions characterised by the reaction format or use of a specific feature
    • C12Q2537/10—Reactions characterised by the reaction format or use of a specific feature the purpose or use of
    • C12Q2537/16—Assays for determining copy number or wherein the copy number is of special importance
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2537/00—Reactions characterised by the reaction format or use of a specific feature
    • C12Q2537/10—Reactions characterised by the reaction format or use of a specific feature the purpose or use of
    • C12Q2537/165—Mathematical modelling, e.g. logarithm, ratio
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/154—Methylation markers
    • C—CHEMISTRY; METALLURGY
    • C12—BIOCHEMISTRY; BEER; SPIRITS; WINE; VINEGAR; MICROBIOLOGY; ENZYMOLOGY; MUTATION OR GENETIC ENGINEERING
    • C12Q—MEASURING OR TESTING PROCESSES INVOLVING ENZYMES, NUCLEIC ACIDS OR MICROORGANISMS; COMPOSITIONS OR TEST PAPERS THEREFOR; PROCESSES OF PREPARING SUCH COMPOSITIONS; CONDITION-RESPONSIVE CONTROL IN MICROBIOLOGICAL OR ENZYMOLOGICAL PROCESSES
    • C12Q2600/00—Oligonucleotides characterized by their use
    • C12Q2600/156—Polymorphic or mutational markers
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G16B20/10—Ploidy or copy number detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G16B20/20—Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B25/00—ICT specially adapted for hybridisation; ICT specially adapted for gene or protein expression
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G16B30/10—Sequence alignment; Homology search
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B40/00—ICT specially adapted for biostatistics; ICT specially adapted for bioinformatics-related machine learning or data mining, e.g. knowledge discovery or pattern finding
    • G16B40/10—Signal processing, e.g. from mass spectrometry [MS] or from PCR
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H10/00—ICT specially adapted for the handling or processing of patient-related medical or healthcare data
    • G16H10/40—ICT specially adapted for the handling or processing of patient-related medical or healthcare data for data related to laboratory analysis, e.g. patient specimen analysis
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H20/00—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance
    • G16H20/10—ICT specially adapted for therapies or health-improving plans, e.g. for handling prescriptions, for steering therapy or for monitoring patient compliance relating to drugs or medications, e.g. for ensuring correct administration to patients
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/20—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for computer-aided diagnosis, e.g. based on medical expert systems
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16H—HEALTHCARE INFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR THE HANDLING OR PROCESSING OF MEDICAL OR HEALTHCARE DATA
    • G16H50/00—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics
    • G16H50/30—ICT specially adapted for medical diagnosis, medical simulation or medical data mining; ICT specially adapted for detecting, monitoring or modelling epidemics or pandemics for calculating health indices; for individual health risk assessment
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16Z—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS, NOT OTHERWISE PROVIDED FOR
    • G16Z99/00—Subject matter not provided for in other main groups of this subclass

Abstract

Disclosed are methods for determining copy number variation (CNV) of a sequence of interest known or suspected to be associated with a variety of medical conditions. In some embodiments, methods are provided for determining copy number variation of fetuses using maternal samples comprising maternal and fetal cell free DNA. In some embodiments, methods are provided for determining CNVs known or suspected to be associated with a variety of medical conditions. The methods involve determining first and second coverages of sequence tags of the reference genome including the sequence of interest. The first coverages are based on sequence reads of cell-free nucleic acid fragments in a first size domain, and the second coverages are based on sequence reads of cell-free nucleic acid fragments in a second size domain. The methods also involve determining the copy number variation in the sequence of interest using a likelihood ratio calculated from the first coverages and the second coverages. Systems for performing the methods are also disclosed. and fetal cell free DNA. In some embodiments, methods are provided for determining CNVs known or suspected to be associated with a variety of medical conditions. The methods involve determining first and second coverages of sequence tags of the reference genome including the sequence of interest. The first coverages are based on sequence reads of cell-free nucleic acid fragments in a first size domain, and the second coverages are based on sequence reads of cell-free nucleic acid fragments in a second size domain. The methods also involve determining the copy number variation in the sequence of interest using a likelihood ratio calculated from the first coverages and the second coverages. Systems for performing the methods are also disclosed.

Description

USING CELL-FREE DNA FRAGMENT SIZE TO INE COPY NUMBER VARIATIONS CROSS REFERENCE TO RELATED APPLICATIONS This application claims benefits under 35 U.S.C. § l 19(e) to U.S.
Provisional Patent Application No. 62/290,891, entitled: USING CELL-FREE DNA FRAGMENT SIZE TO DETERMINE COPY NUMBER VARIATIONS, filed February 3, 2016, and U.S. Patent Application No. 15/382,508, entitled: USING CELL-FREE DNA FRAGMENT SIZE TO DETERMINE COPY NUMBER VARIATIONS, filed 16 er 2016, which are herein incorporated by reference in their entireties for all purposes.
BACKGROUND One of the critical endeavors in human medical research is the discovery of genetic abnormalities that produce adverse health consequences. In many cases, specific genes and/or al diagnostic markers have been fied in portions of the genome that are present at al copy numbers. For example, in prenatal diagnosis, extra or missing copies of whole chromosomes are frequently occurring genetic lesions. In cancer, deletion or multiplication of copies of whole chromosomes or chromosomal segments, and higher level amplifications of ic regions ofthe genome, are common occurrences.
[0002] Most information about copy number variation (CNV) has been provided by cytogenetic resolution that has permitted recognition of structural abnormalities. Conventional procedures for genetic screening and biological dosimetry have utilized invasive ures, e.g., amniocentesis, cordocentesis, or chorionic villus sampling (CVS), to obtain cells for the analysis of karyotypes.
Recognizing the need for more rapid testing methods that do not require cell culture, fluorescence in situ hybridization (FISH), quantitative fluorescence PCR (QF-PCR) and array- Comparative c Hybridization (array-CGH) have been developed as molecular-cytogenetic methods for the analysis of copy number variations.
One of the critical endeavors in human medical research 1s the discovery of genetic alities that produce adverse health consequences. In many cases, specific genes and/or critical diagnostic markers have been identified in portions of the genome that are present at abnormal copy numbers. For example, in prenatal diagnosis, extra or missing copies of whole chromosomes are frequently occurring genetic lesions. In cancer, deletion or multiplication of copies of whole chromosomes or chromosomal segments, and higher level amplifications of specific regions ofthe genome, are common occurrences.
Most information about copy number variation (CNV) has been provided by cytogenetic tion that has permitted recognition of structural abnormalities. tional procedures for genetic screening and biological I0 try have utilized invasive procedures, e.g., amniocentesis, cordocentesis, or chorionic villus sampling (CVS), to obtain cells for the analysis of karyotypes.
Recognizing the need for more rapid testing s that do not require cell culture, fluorescence in situ hybridization , quantitative scence PCR (QF-PCR) and array- Comparative Genomic Hybridization (array-CGH) have been developed as molecular-cytogenetic methods for the analysis of copy number variations.
The advent oftechnologies that allow for sequencing entire genomes in relatively short time, and the discovery of circulating cell-free DNA (cIDNA) have provided the opportunity to compare genetic material originating from one chromosome to be compared to that of another without the risks associated with invasive ng methods, which provides a tool to diagnose various kinds of copy number variations tic sequences ofinterest.
Limitations of existing methods in noninvasive prenatal diagnostics, which include icient sensitivity stemming from the limited levels of cIDNA, and the sequencing bias of the logy stemming from the inherent nature of genomic information, ie the continuing need for noninvasive methods that would provide any or all of the specificity, sensitivity, and applicability, to reliably diagnose copy number changes in a variety of clinical settings. It has been shown that the average lengths of the fetal cIDNA fragments are shorter than the maternal cIDNA fragments in the plasma of pregnant women. This difference between maternal and fetal cIDNA is exploited in the implementation herein to determine CNV and/or fetal on.
Embodiments sed herein fulfill some of the above needs. Some embodiments may be implemented with a PCR free library preparation coupled with paired end DNA sequencing. Some embodiments provide high analytical sensitivity and specificity for noninvasive prenatal diagnostics and diagnoses of a variety of diseases.
SUMMARY In some embodiments, methods are provided for determining copy number variation (CNV) of any fetal aneuploidy, and CNVs known or suspected to be associated with a variety of medical ions. CNVs that can be determined according to the t method include trisomies and monosomies of any one or more of chromosomes 1-22, X and Y, other chromosomal polysomies, and deletions and/or duplications of ts of any one or more of the chromosomes. In some embodiments, the methods involve identifying CNVs of a nucleic acid sequence of interest, e.g., a clinically relevant sequence, in a test . The method assesses copy number variation ofthe specific sequence ofinterest.
In some embodiments, the method is implemented at a computer system that includes one or more sors and system memory to te copy number of a nucleic acid sequence of interest in a test sample comprising nucleic acids of one or more genomes.
One aspect ofthe disclosure relates to a method for determining a copy number variation (CNV) of a nucleic acid sequence of st in a test sample including cell-free nucleic acid fragments originating from two or more genomes.
The method includes: (a) receiving sequence reads obtained by sequencing the cellfree nucleic acid fragments in the test sample; (b) aligning the sequence reads of the cell-free nucleic acid nts or aligning fragments containing the sequence reads to bins of a nce genome including the ce of interest, thereby providing test sequence tags, wherein the reference genome is divided into a plurality of bins; (c) determining fragment sizes of at least some of the cell-free nucleic acid fragments present in the test sample; (d) calculating coverages of the sequence tags for the bins of the reference genome by, for each bin: (i) determining a number of sequence tags aligning to the bin, and (ii) izing the number of sequence tags ng to the bin by accounting for bin-to-bin variations due to s other than copy number variation; (e) determining at-statistic for the sequence of st using coverages of bins in the sequence of interest and coverages of bins in a reference region for the sequence of interest; and (f) determining a copy number variation in the sequence of interest using a likelihood ratio calculated from the t-statistic and information about the sizes of the cell-free nucleic acid fragments.
In some entations, the method includes performing (d) and (e) twice, once for fragments in a first size domain and again for fragments in a second size domain. In some implementations, the first size domain es cell-free nucleic acid fragments of substantially all sizes in the sample, and the second size domain includes only cell-free nucleic acid fragments r than a defined size. In some implementations, the second size domain includes only the cell-free nucleic acid fragments r than about 150 bp. In some implementations, the likelihood ratio is calculated from a first t-statistic for the sequence of interest using sequence tags for fragments in a first size range, and a second t-statistic for the sequence of interest using sequence tags for fragments in a second size range.
In some implementations, the likelihood ratio is calculated as a first likelihood that the test sample is an aneuploid sample over a second likelihood that the test sample is a euploid sample.
In some implementations, the likelihood ratio is ated from one or more values of fetal fraction in addition to the t-statistic and information about the sizes ofthe cell-free nucleic acid fragments.
In some implementations, the one or more values of fetal fraction include a value of fetal fraction calculated using the information about the sizes ofthe cell-free nucleic acid fragments. In some implementations, the value of fetal fraction is calculated by: obtaining a frequency bution of the fragment sizes; and applying the frequency distribution to a model relating fetal fraction to frequency of fragment size to obtain the fetal fraction value. In some implementations, the model relating fetal fraction to frequency of fragment size includes a l linear model having a plurality s and coefficients for a plurality offragment sizes.
In some implementations, the one or more values of fetal fraction include a value of fetal fraction calculated using coverage information for the bins of the reference genome. In some implementations, the value of fetal on is calculated by applying coverage values of a plurality of bins to a model relating fetal fraction to ge of bin to obtain the fetal fraction value. In some entations, the model relating fetal fraction to the coverage of bin includes a general linear model having a plurality of terms and cients for a plurality of bins. In some implementations, the plurality of bins have high ation between fetal fraction and coverage in training samples.
In some entations, the one or more values of fetal fraction include a value of fetal fraction calculated using frequencies of a plurality of 8-mers found in the reads. In some implementations, the value of fetal fraction is calculated by: applying frequencies of a plurality of 8-mers to a model relating fetal fraction to 8-mer frequency to obtain the fetal fraction value. In some implementations, the model relating fetal on to 8-mer frequency includes a general linear model I0 having a plurality of terms and coefficients for a plurality of 8-mers. In some implementations, the plurality of 8-mers have high correlation between fetal fraction and 8-mer frequency.
In some implementations, the one or more values of fetal fraction include a value of fetal fraction calculated using coverage ation for the bins of a sex chromosome.
In some implementations, the likelihood ratio is calculated from a fetal fraction, at-statistic of short fragments, and at statistics of all fragments, wherein the short fragments are cell-free nucleic acid fragments in a first size range smaller than a criterion size, and the all nts are ree nucleic acid fragments including the short fragments and fragments longer than the criterion size. In some implementations, the likelihood ratio is calculated: LR = al q(fftota1)*P1(Tshort•Ta11lffest) Po (Tshort,Tall) where p 1 represents the likelihood that data come from a multivariate normal distribution representing a 3-copy or I-copy model, p 0 represents the likelihood that data come from a multivariate normal distribution representing a 2- copy model, Tshort, Tan are T scores ated from chromosomal coverage generated from short fragments and all fragments, and q(fftotaz) is a y distribution of the fetal fraction.
In some entations, the likelihood ratio is calculated from one or more values of fetal fraction in addition to the t-statistic and information about the sizes ofthe cell-free nucleic acid fragments.
In some implementations, the likelihood ratio is calculated for monosomy X, trisomy X, trisomy 13, trisomy 18, or trisomy 21.
In some implementations, normalizing the number of sequence tags includes: normalizing for GC t of the sample, normalizing for a global wave profile of variation of a training set, and/or normalizing for one or more components obtained from a principal component analysis.
In some implementations, the sequence of interest is a human chromosome selected from the group consisting of some 13, chromosome 18, chromosome 21, chromosome X, and chromosome Y.
[0024] In some implementations, the reference reg10n 1s all robust chromosomes, robust chromosomes not including the sequence of interest, at least a some outside of the sequence of interest, and/or a subset of chromosomes ed from the robust chromosomes. In some implementations, the nce region includes robust chromosomes that have been determined to e the best signal detection ability for a set oftraining samples.
In some implementations, the method further includes calculating values of a size parameter for the bins by, for each bin: (i) determining a value of the size parameter from sizes of cell-free nucleic acid fragments in the bin, and (ii) normalizing the value of the size parameter by accounting for bin-to-bin variations due to factors other than copy number variation. The method also includes determining a size-based t-statistic for the ce of interest using values ofthe size parameter of bins in the ce of interest and values of the size parameter of bins in the nce region for the sequence of interest. In some implementations, the likelihood ratio of (f) is calculated from the t-statistic and the ased t-statistic. In some entations, the likelihood ratio of (f) is calculated from the size-based tstatistic and a fetal fraction.
In some implementations, the method r includes comparing the likelihood ratio to a call criterion to determine a copy number variation in the sequence of interest. In some implementations, the likelihood ratio is converted to a log likelihood ratio before being compared to the call criterion. In some implementations, the call criterion is obtained by applying different criteria to a training set of training samples, and selecting a criterion that provides a defined sensitivity and a defined selectivity.
In some implementations, the method further includes obtaining a plurality of likelihood ratios and ng the ity of hood ratios to a decision tree to determine a ploidy case for the sample.
In some implementations, the method further includes obtaining a ity of likelihood ratios and one or more coverage values of the sequence of interest, and applying the plurality of likelihood ratios and one or more coverage values of the sequence of interest to a decision tree to determine a ploidy case for the sample.
Another aspect of the disclosure s to a method for determining a copy number variation (CNV) of a nucleic acid sequence of interest in a test sample including cell-free nucleic acid fragments originating from two or more genomes.
The method includes: (a) receiving sequence reads obtained by sequencing the cell- free nucleic acid fragments in the test sample; (b) aligning the sequence reads of the cell-free nucleic acid fragments or aligning fragments containing the sequence reads to bins of a reference genome including the ce of interest, thereby providing test sequence tags, wherein the reference genome is d into a plurality of bins; (c) calculating coverages ofthe sequence tags for the bins ofthe reference genome by, for each bin: (i) ining a number of sequence tags aligning to the bin, and (ii) normalizing the number of sequence tags aligning to the bin by accounting for bin-tobin variations due to factors other than copy number variation. The method also includes: (d) determining at-statistic for the sequence of interest using coverages of bins in the sequence of interest and coverages of bins in a nce region for the sequence of interest; (e) estimating one or more fetal fraction values of the cell-free nucleic acid fragments in the test sample; and (f) ining a copy number variation in the sequence of st using the t-statistic and the one or more fetal fraction values.
In some implementations, (f) includes calculating a likelihood ratio from the t-statistic and the one or more fetal fraction values. In some implementations, the hood ratio is calculated for monosomy X, trisomy X, trisomy 13, trisomy 18, or trisomy 21.
In some implementations, normalizing the number of ce tags includes: normalizing for GC content of the sample, normalizing for a global wave profile of variation of a training set, and/or normalizing for one or more components obtained from a principal component is.
[0032] In some implementations, the sequence of interest is a human chromosome selected from the group consisting of chromosome 13, chromosome 18, chromosome 21, chromosome X, and chromosome Y.
A further aspect of the disclosure relates to a method for determining a copy number ion (CNV) of a nucleic acid sequence of interest in a test sample including cell-free nucleic acid fragments originating from two or more s.
The method includes: (a) receiving sequence reads obtained by sequencing the cellfree nucleic acid fragments in the test ; (b) aligning the sequence reads of the cell-free nucleic acid fragments or aligning fragments ning the sequence reads to bins of a reference genome including the ce of interest, thereby providing test sequence tags, wherein the nce genome is divided into a plurality of bins; (c) determining fragment sizes of the cell-free nucleic acid fragments existing in the test sample; (d) calculating coverages of the sequence tags for the bins of the nce genome using sequence tags for the ree nucleic acid fragments having sizes in a first size domain; (e) calculating coverages of the sequence tags for the bins of the reference genome using sequence tags for the cell-free nucleic acid fragments having sizes in a second size domain, wherein the second size domain is different from the first size domain; (f) calculating size characteristics for the bins of the reference genome using the fragment sizes ined in (c); and (g) determining a copy number variation in the ce of interest using the coverages calculated in (d) and (e) and the size characteristics calculated in (f).
In some implementations, the first size domain includes cell-free c acid fragments of substantially all sizes in the sample, and the second size domain includes only cell-free nucleic acid fragments smaller than a defined size. In some implementations, the second size domain includes only the cell-free nucleic acid fragments smaller than about 150 bp.
In some implementations, the sequence of interest is a human chromosome selected from the group consisting of chromosome 13, chromosome 18, chromosome 21, chromosome X, and chromosome Y.
In some implementations, (g) includes calculating a t-statistic for the sequence of interest using the coverages of bins in the sequence of interest calculated in (d) and/or (e). In some implementations, wherein calculating the t-statistic for the sequence of st includes using the coverages of bins in the sequence of interest and coverages ofbins in a reference region for the sequence of interest.
In some implementations, (g) includes calculating a t-statistic for the sequence of interest using the size characteristics of bins in the sequence of interest calculated in (f). In some implementations, calculating the t-statistic for the sequence of interest includes using the size characteristics of bins in the ce of interest and size characteristics ofbins in a reference region for the sequence of interest.
In some implementations, the size characteristic for a bin includes a ratio ments of size smaller than a d value to total fragments in the bin.
In some implementations, (g) includes calculating a likelihood ratio from the t-statistic.
In some implementations, (g) includes calculating a likelihood ratio from a first istic for the sequence of interest using the coverages calculated in (d), and a second t-statistic for the ce of interest using the coverages calculated in (e).
In some implementations, (g) includes calculating a likelihood ratio from a first t-statistic for the sequence of interest using the coverages calculated in (d), a second t-statistic for the ce of interest using the coverages calculated in (e), and third t-statistic for the ce of interest using the size characteristics ated in (f).
In some implementations, the likelihood ratio is calculated from one or more values of fetal fraction in on to at least the first and second istic. In some implementations, the method further includes calculating the one or more values of fetal fraction using the information about the sizes of the cell-free c acid fragments.
In some implementations, the method r includes calculating the one or more values of fetal fraction using coverage information for the bins of the reference genome. In some implementations, the one or more values of fetal fraction include a value of fetal fraction calculated using ge information for the bins of a sex chromosome. In some implementations, the hood ratio is calculated for monosomy X, trisomy X, trisomy 13, y 18, or trisomy 21.
In some implementations, (d) and/or (e) includes: (i) determining a number of ce tags aligning to the bin, and (ii) normalizing the number of sequence tags aligning to the bin by accounting for bin-to-bin variations due to factors other than copy number variation. In some implementations, izing the number of sequence tags includes: normalizing for GC content of the sample, normalizing for a global wave profile ofvariation of a training set, and/or normalizing for one or more components obtained from a principal component is.
In some implementations, (f) includes calculating values of a size parameter for the bins by, for each bin: (i) ining a value of the size parameter from sizes of cell-free nucleic acid fragments in the bin, and (ii) normalizing the value of the size parameter by accounting for bin-to-bin variations due to factors other than copy number variation.
Another aspect of the disclosure relates to a system for evaluation of copy number of a nucleic acid sequence of interest in a test , the system includes: a sequencer for receiving nucleic acid fragments from the test sample and providing nucleic acid sequence information of the test sample; a processor; and one or more computer-readable storage media having stored thereon instructions for execution on said processor. The instructions includes instruction to: (a) receive sequence reads obtained by sequencing the cell-free nucleic acid fragments in the test sample; (b) align the ce reads ofthe cell-free nucleic acid nts or aligning fragments containing the sequence reads to bins of a reference genome including the sequence of interest, y providing test sequence tags, wherein the reference genome is divided into a plurality of bins; (c) determine fragment sizes of at least some of the cell-free nucleic acid fragments present in the test sample; and (d) calculate coverages of the sequence tags for the bins of the reference genome by, for each bin: (i) determining a number of sequence tags aligning to the bin, and (ii) normalizing the number of sequence tags aligning to the bin by accounting for bin-to- bin variations due to factors other than copy number variation. The method also includes: (e) determine tistic for the sequence ofinterest using coverages of bins in the sequence of interest and coverages of bins in a reference region for the sequence of interest; and (f) determine a copy number variation in the sequence of interest using a likelihood ratio calculated from the t-statistic and information about the sizes of the cell-free c acid fragments.
In some implementations, the system is configured to perform any of the methods described above.
An additional aspect of the sure relates a computer program product ing one or more computer-readable non-transitory storage media having stored thereon computer-executable instructions that, when executed by one or more processors of a computer system, cause the computer system to implement any ofthe methods above. gh the examples herein concern humans and the language is primarily directed to human concerns, the concepts described herein are applicable to genomes from any plant or animal. These and other objects and features of the present disclosure will become more fully apparent from the following description and ed claims, or may be learned by the practice of the disclosure as set forth hereinafter.
INCORPORATION BY REFERENCE All patents, patent applications, and other publications, ing all sequences disclosed within these references, referred to herein are expressly incorporated herein by reference, to the same extent as if each individual publication, patent or patent application was specifically and individually ted to be incorporated by reference. All nts cited are, in relevant part, incorporated herein by reference in their entireties for the purposes indicated by the context of their citation herein. However, the citation of any document is not to be construed as an ion that it is prior art with respect to the present disclosure.
BRIEF PTION OF THE DRAWINGS Figure 1 is a flowchart of a method 100 for determining the presence or absence of a copy number variation in a test sample comprising a mixture of nucleic acids.
[0052] Figure 2A thematically illustrates how paired end sequencing may be used to determine both fragment size and sequence coverage.
Figure 2B shows a flowchart of a process for usmg size-based coverage to determine a copy number variation of a nucleic acid ce of interest in a test sample.
[0054] Figure 2C depicts a flowchart of a process for determining fragment size ter for a nucleic acid sequence of interest used for evaluation of the copy number.
Figure 2D shows a flow chart oftwo overlapping passes ofworkflow.
Figure 2E shows a flow chart of a three-pass process for evaluating copy number.
Figure 2F shows implementations that apply a t-statistic to copy number analysis to improve the accuracy ofthe analysis.
Figure 2G shows an example process for determining fetal fraction from ge information ing to some implementations ofthe disclosure.
[0059] Figure 2H shows a process for determining fetal fraction from size distribution information according to some implementations.
Figure 21 shows an example process for determining fetal fraction from 8-mer frequency information according to some implementations ofthe disclosure.
Figure 2J shows a ow for processing ce reads information ofwhich can be used to obtain fetal fraction estimates.
Figure 3A shows a flowchart of an example of a process for ng the noise in sequence data from a test sample.
Figures 3B-3K present analyses of data obtained at various stages of the process depicted in Figure 3A. 2016/067886 Figure 4A shows a flow chart of a process for ng a sequence mask for reducing noise in sequence data.
Figure 4B shows that MapQ score has a strong monotonous correlation with CV ofnormalized ge quantities.
[0066] Figure 5 is a block diagram of a dispersed system for processing a test sample and tely making a diagnosis.
Figure 6 schematically rates how different operations m processing test samples may be grouped to be handled by different elements of a system.
[0068] s 7A and 7B shows electropherograms of a cIDNA sequencing library prepared according to the abbreviated protocol described in Example la (Fig. 7A), and the protocol described in Example I b (Fig. 7B ).
Figure 8 shows the overall workflow and ne for a new version of NIPT compared to the standard laboratory ow.
[0070] Figure 9 shows sequencing library yield as a function of input extracted cIDNA, indicating a strong linear correlation with library concentration to input concentration with a high conversion efficiency.
Figure I 0 shows the cIDNA fragment size distribution as measured from 324 samples from pregnancies with a male fetus.
[0072] Figure 11 shows the relative fetal fraction from the total counts of mapped paired end reads compared to the counts from paired end reads that are less than 150 bp.
Figure 12 shows ed istic aneuploidy score for detection of trisomy 21 samples for (A) counts of all fragments; (B) counts of short fragments (<150bp) only; (C) fraction of short fragments (counts between 80 and 150 bp/counts <250bp); (D) combined t-statistic from (B) and (C); and (E) results for same samples obtained using the Illumina Redwood City CLIA laboratory process with an average of 16 M counts/sample.
Figure 13 shows fetal fractions estimated from selected bins versus those measured with normalized chromosome values (REF) for the X-chromosome. Set I was used to calibrate the fetal fraction value and an independent set 2 to test the correlation.
DETAILED DESCRIPTION Definitions
[0075] Unless ise indicated, the practice of the method and system disclosed herein es conventional techniques and apparatus commonly used in molecular biology, microbiology, protein purification, protein engineering, protein and DNA sequencing, and recombinant DNA fields, which are within the skill of the art. Such techniques and tus are known to those of skill in the art and are described in numerous texts and reference works (See e.g., Sambrook et al., "Molecular Cloning: A Laboratory Manual," Third Edition (Cold Spring Harbor), ); and Ausubel et al., "Current Protocols in lar y" [1987]).
Numeric ranges are inclusive of the numbers defining the range. It is intended that every maximum numerical limitation given throughout this specification includes every lower cal limitation, as if such lower numerical limitations were expressly written herein. Every minimum numerical limitation given throughout this specification will include every higher numerical tion, as if such higher numerical limitations were expressly written herein. Every numerical range given hout this specification will include every narrower numerical range that falls within such broader numerical range, as if such narrower numerical ranges were all expressly written herein.
The headings provided herein are not intended to limit the disclosure.
Unless defined otherwise herein, all technical and scientific terms used herein have the same meaning as commonly tood by one of ordinary skill in the art. Various scientific dictionaries that include the terms ed herein are well known and ble to those in the art. Although any methods and materials similar or equivalent to those bed herein find use in the practice or testing of the embodiments disclosed herein, some methods and materials are described.
The terms defined immediately below are more fully described by reference to the Specification as a whole. It is to be understood that this disclosure is not limited to the particular methodology, ols, and reagents described, as these may vary, depending upon the context they are used by those of skill in the art. As used herein, the singular terms "a," "an," and "the" include the plural reference unless the context clearly indicates otherwise.
Unless otherwise indicated, nucleic acids are written left to right in 5' to 3' orientation and amino acid sequences are n left to right in amino to carboxy orientation, respectively.
The term "parameter" 1s used herein represents a physical feature whose value or other characteristic has an impact a relevant condition such as copy number ion. In some cases, the term parameter is used with reference to a variable that affects the output of a mathematical relation or model, which le may be an independent variable (i.e., an input to the model) or an intermediate variable based on one or more independent variables. Depending on the scope of a model, an output of one model may become an input of another model, thereby becoming a parameter to the other model.
[0082] The term "fragment size ter" refers to a parameter that relates to the size or length of a fragment or a collection of fragments such nucleic acid fragments; e.g., a cIDNA nts obtained from a bodily fluid. As used herein, a parameter is d toward a fragment size or size range" when: 1) the parameter is favorably weighted for the fragment size or size range, e.g., a count weighted more y when associated with fragments of the size or size range than for other sizes or ranges; or 2) the parameter is obtained from a value that is favorably weighted for the fragment size or size range, e.g., a ratio obtained from a count weighted more heavily when associated with nts of the size or size range. A fragment size or size range may be a characteristic of a genome or a portion thereof when the genome produces nucleic acid fragments enriched in or having a higher tration of the size or size range relative to nucleic acid fragments from another genome or another portion ofthe same genome.
The term "weighting" refers to modifying a quantity such as a parameter or variable using one or more values or functions, which are considered the t." In certain embodiments, the ter or variable is multiplied by the weight. In other embodiments, the parameter or variable is ed exponentially. In some embodiments, the function may be a linear or non-linear function. Examples of 2016/067886 applicable non-linear functions include, but are not limited to Heaviside step functions, box-car functions, stair-case functions, or sigmoidal functions. Weighting an original parameter or variable may systematically increase or decrease the value of the weighted variable. In various embodiments, weighting may result in ve, non- negative, or negative values.
The term "copy number variation" herein refers to variation in the number of copies of a nucleic acid sequence present in a test sample in comparison with the copy number of the nucleic acid ce present in a reference sample. In certain embodiments, the nucleic acid sequence is 1 kb or . In some cases, the nucleic acid sequence is a whole chromosome or significant portion thereof. A "copy number variant" refers to the sequence of c acid in which copy-number differences are found by comparison of a nucleic acid ce of interest in test sample with an expected level of the nucleic acid sequence of interest. For example, the level ofthe nucleic acid sequence of interest in the test sample is compared to that t in a qualified sample. Copy number variants/variations include deletions, ing microdeletions, insertions, including microinsertions, duplications, multiplications, and ocations. CNVs encompass chromosomal aneuploidies and partial aneuploidies.
The term loidy" herein refers to an imbalance of genetic material caused by a loss or gain of a whole some, or part of a some.
The terms "chromosomal aneuploidy" and "complete chromosomal aneuploidy" herein refer to an imbalance of genetic material caused by a loss or gain of a whole chromosome, and includes germline aneuploidy and mosaic aneuploidy.
The terms "partial aneuploidy" and "partial chromosomal aneuploidy" herein refer to an nce of genetic material caused by a loss or gain of part of a chromosome, e.g., partial monosomy and partial trisomy, and encompasses imbalances resulting from translocations, deletions and insertions.
The term "plurality" refers to more than one element. For example, the term is used herein in reference to a number of nucleic acid molecules or sequence tags that are sufficient to identify significant differences in copy number variations in test s and qualified samples using the methods disclosed herein. In some embodiments, at least about 3 x 106 sequence tags of between about 20 and 40bp are obtained for each test sample. In some embodiments, each test sample provides data for at least about 5 x 106, 8 x 106, 10 x 106, 15 x 106, 20 x 106, 30 x 106, 40 x 106, or 50 x 106 sequence tags, each sequence tag comprising between about 20 and 40bp.
The term "paired end reads" refers to reads from paired end sequencing that obtains one read from each end of a nucleic acid fragment. Paired end sequencing may involve nting strands of polynucleotides into short sequences called inserts. Fragmentation is optional or unnecessary for relatively short polynucleotides such as cell free DNA les.
The terms "polynucleotide," "nucleic acid" and "nucleic acid molecules" are used interchangeably and refer to a covalently linked ce of nucleotides (i.e., ribonucleotides for RNA and deoxyribonucleotides for DNA) in which the 3' position of the pentose of one nucleotide is joined by a phosphodiester group to the 5' position ofthe pentose ofthe next. The nucleotides include sequences of any form of c acid, including, but not limited to RNA and DNA molecules such as cIDNA molecules. The term "polynucleotide" includes, without limitation, single- and double-stranded polynucleotide.
The term "test sample" herein refers to a sample, typically derived from a biological fluid, cell, tissue, organ, or organism, comprising a nucleic acid or a mixture of nucleic acids comprising at least one nucleic acid sequence that is to be screened for copy number variation. In certain embodiments the sample ses at least one c acid sequence whose copy number is suspected of having undergone variation. Such samples e, but are not limited to sputum/oral fluid, amniotic fluid, blood, a blood on, or fine needle biopsy samples (e.g., surgical biopsy, fine needle biopsy, etc.), urine, peritoneal fluid, pleural fluid, and the like. Although the sample is often taken from a human subject (e.g., patient), the assays can be used to copy number variations (CNVs) in samples from any mammal, including, but not limited to dogs, cats, horses, goats, sheep, , pigs, etc. The sample may be used directly as obtained from the biological source or following a pretreatment to modify the ter of the sample. For example, such pretreatment may e preparing plasma from blood, diluting viscous fluids and so forth. Methods of pretreatment may also involve, but are not limited to, filtration, precipitation, dilution, distillation, mixing, centrifugation, ng, lyophilization, concentration, amplification, nucleic acid fragmentation, inactivation of interfering components, the on of reagents, lysing, etc. Ifsuch methods of pretreatment are employed with respect to the sample, such pretreatment methods are typically such that the nucleic acid(s) of interest remain in the test sample, sometimes at a concentration proportional to that in an untreated test sample (e.g., namely, a sample that is not subjected to any such pretreatment method(s)). Such "treated" or "processed" samples are still considered to be biological "test" samples with respect to the methods described herein.
The term fied " or "unaffected sample" herein refers to a sample comprising a mixture of c acids that are present in a known copy number to which the c acids in a test sample are to be compared, and it is a sample that is normal, i.e., not aneuploid, for the nucleic acid ce of interest. In some ments, qualified samples are used as unaffected training samples of a training set to derive sequence masks or sequence profiles. In certain embodiments, qualified samples are used for identifying one or more normalizing chromosomes or segments for a chromosome under consideration. For example, qualified samples may be used for fying a normalizing chromosome for chromosome 21. In such case, the ied sample is a sample that is not a trisomy 21 sample. Another e es using only females as ying samples for chromosome X.
Qualified samples may also be employed for other purposes such as determining thresholds for calling affected samples, identifying thresholds for defining mask regions on a reference sequence, determining expected coverage quantities for different regions of a genome, and the like.
The term "training set" herein refers to a set of training samples that can comprise affected and/or unaffected samples and are used to develop a model for analyzing test samples. In some embodiments, the ng set includes unaffected samples. In these embodiments, thresholds for determining CNV are established using training sets of samples that are unaffected for the copy number variation of interest. The unaffected samples in a training set may be used as the qualified samples to fy normalizing sequences, e.g., normalizing chromosomes, and the chromosome doses ofunaffected samples are used to set the thresholds for each ofthe sequences, e.g., chromosomes, of interest. In some embodiments, the training set includes affected samples. The affected s in a training set can be used to verify that affected test samples can be easily entiated from unaffected samples.
A training set is also a statistical sample in a population of interest, which statistical sample is not to be confused with a ical . A tical sample often ses multiple individuals, data of which individuals are used to ine one or more quantitative values of interest generalizable to the population.
The statistical sample is a subset of duals in the population of interest. The individuals may be s, animals, tissues, cells, other ical samples (i.e., a statistical sample may include multiple biological samples), and other individual es providing data points for statistical analysis.
Usually, a training set is used in ction with a validation set. The term "validation set" is used to refer to a set of individuals in a statistical sample, data ofwhich individuals are used to validate or evaluate the quantitative values ofinterest determined using a training set. In some embodiments, for instance, a training set provides data for calculating a mask for a reference sequence, while a validation set provides data to evaluate the validity or effectiveness ofthe mask.
[0096] "Evaluation of copy " is used herein in reference to the statistical evaluation of the status of a genetic sequence related to the copy number of the sequence. For example, in some embodiments, the evaluation comprises the determination of the presence or absence of a genetic sequence. In some embodiments the evaluation comprises the determination of the partial or complete aneuploidy of a genetic sequence. In other embodiments the evaluation ses discrimination n two or more samples based on the copy number of a genetic sequence. In some embodiments, the evaluation comprises statistical analyses, e.g., normalization and comparison, based on the copy number ofthe genetic sequence.
The term "qualified nucleic acid" is used interchangeably with "qualified sequence," which is a sequence against which the amount of a sequence or nucleic acid of interest is compared. A qualified sequence is one present in a biological sample preferably at a known representation, i.e., the amount of a qualified sequence is known. Generally, a qualified sequence is the sequence present in a "qualified sample." A "qualified sequence of interest" is a qualified sequence for which the amount is known in a qualified sample, and is a sequence that is associated with a difference of a sequence of st between a control subject and an dual with a medical condition.
The term "sequence of interest" or "nucleic acid ce of interest" herein refers to a c acid sequence that is associated with a ence in ce representation between healthy and diseased individuals. A sequence of interest can be a sequence on a chromosome that is misrepresented, i.e., over- or under-represented, in a disease or genetic condition. A ce of interest may be a portion of a chromosome, i.e., chromosome segment, or a whole chromosome. For example, a sequence of interest can be a chromosome that is over-represented in an oidy condition, or a gene encoding a tumor-suppressor that is underrepresented in a cancer. Sequences of interest include sequences that are over- or under- represented in the total population, or a ulation of cells of a subject. A "qualified sequence ofinterest" is a sequence ofinterest in a qualified sample. A "test ce ofinterest" is a sequence ofinterest in a test sample.
The term "normalizing sequence" herein refers to a sequence that is used to normalize the number of ce tags mapped to a sequence of interest associated with the izing sequence. In some embodiments, a normalizing sequence comprises a robust chromosome. A "robust chromosome" is one that is unlikely to be aneuploid. In some cases involving the human chromosome, a robust chromosome is any chromosome other than the X chromosome, Y chromosome, chromosome 13, chromosome 18, and chromosome 21. In some embodiments, the izing sequence displays a variability in the number of sequence tags that are mapped to it among samples and sequencing runs that approximates the variability of the sequence of interest for which it is used as a normalizing parameter. The normalizing sequence can differentiate an affected sample from one or more unaffected s. In some implementations, the normalizing sequence best or effectively differentiates, when compared to other potential normalizing sequences such as other chromosomes, an affected sample from one or more unaffected samples.
In some embodiments, the variability of the normalizing ce is calculated as the variability in the chromosome dose for the sequence of interest across samples and sequencing runs. In some embodiments, normalizing sequences are identified in a set ofunaffected samples.
A "normalizing chromosome," "normalizing nator chromosome," or "normalizing chromosome sequence" is an example of a "normalizing sequence." A lizing chromosome sequence" can be composed of a single chromosome or of a group of chromosomes. In some embodiments, a normalizing sequence comprises two or more robust chromosomes. In certain embodiments, the robust chromosomes are all autosomal chromosomes other than chromosomes, X, Y, 13, 18, and 21. A "normalizing t" is another example of a lizing sequence." A "normalizing segment sequence" can be composed of a single segment of a some or it can be composed of two or more segments of the same or of ent chromosomes. In certain ments, a normalizing sequence 1s intended to normalize for variability such as process-related, interchromosomal (intra-run), and inter-sequencing (inter-run) variability.
[00101] The term "differentiability" herein refers to a teristic of a normalizing chromosome that enables one to distinguish one or more unaffected, i.e., normal, samples from one or more affected, i.e., aneuploid, samples. A normalizing chromosome displaying the greatest "differentiability" is a chromosome or group of somes that provides the st statistical difference between the distribution of chromosome doses for a chromosome of interest in a set of qualified samples and the chromosome dose for the same chromosome of interest in the corresponding chromosome in the one or more affected samples.
The term "variability" herein refers to another characteristic of a normalizing chromosome that enables one to guish one or more unaffected, i.e., normal, samples from one or more affected, i.e., aneuploid, samples. The variability of a normalizing chromosome, which is measured in a set of qualified samples, refers to the variability in the number of sequence tags that are mapped to it that approximates the variability in the number of sequence tags that are mapped to a chromosome ofinterest for which it serves as a normalizing ter.
[00103] The term "sequence tag density" herein refers to the number of sequence reads that are mapped to a reference genome sequence, e.g., the sequence tag density for chromosome 21 is the number of ce reads generated by the cing method that are mapped to chromosome 21 ofthe reference .
The term "sequence tag density ratio" herein refers to the ratio of the number of sequence tags that are mapped to a chromosome of the reference genome, e.g., chromosome 21, to the length ofthe reference genome chromosome.
The term "sequence dose" herein refers to a parameter that relates the number of sequence tags or another parameter identified for a sequence of st and the number of sequence tags or the other parameter identified for the izing sequence. In some cases, the sequence dose is the ratio of the ce tag coverage or the other parameter for a sequence of interest to the sequence tag coverage or the other parameter for a normalizing sequence. In some cases, the sequence dose refers to a parameter that relates the sequence tag density of a sequence of interest to the sequence tag density of a normalizing sequence. A "test sequence dose" is a parameter that relates the sequence tag y or the other parameter of a sequence of interest, e.g., chromosome 21, to that of a normalizing sequence, e.g., chromosome 9, determined in a test sample. Similarly, a "qualified sequence dose" is a parameter that relates the sequence tag density or the other parameter of a sequence of interest to that of a normalizing sequence determined in a qualified sample.
The term "coverage" refers to the abundance of sequence tags mapped to a defined ce. Coverage can be quantitatively ted by sequence tag density (or count of sequence tags), ce tag density ratio, normalized ge amount, adjusted coverage values, etc.
The term "coverage quantity" refers to a modification of raw coverage and often represents the relative quantity of sequence tags (sometimes called counts) in a region of a genome such as a bin. A coverage quantity may be obtained by normalizing, adjusting and/or correcting the raw coverage or count for a region ofthe genome. For example, a normalized coverage quantity for a region may be obtained by dividing the sequence tag count mapped to the region by the total number sequence tags mapped to the entire genome. ized ge quantity allows comparison of coverage of a bin across different samples, which may have ent depths of sequencing. It s from sequence dose in that the latter is typically obtained by dividing by the tag count mapped to a subset of the entire genome. The subset is one or more normalizing segments or chromosomes. Coverage quantities, whether or not normalized, may be corrected for global profile variation from region to region on the genome, G-C fraction variations, outliers in robust chromosomes, etc.
The term "Next Generation Sequencing (NGS)" herein refers to cmg methods that allow for massively parallel sequencing of clonally ied les and of single nucleic acid les. Non-limiting examples of WO 36059 NGS include sequencing-by-synthesis usmg reversible dye terminators, and sequencing-by-ligation.
The term "parameter" herein refers to a numerical value that characterizes a property of a system. Frequently, a parameter numerically characterizes a quantitative data set and/or a cal relationship between quantitative data sets. For example, a ratio (or function of a ratio) between the number of sequence tags mapped to a chromosome and the length ofthe chromosome to which the tags are mapped, is a parameter.
The terms "threshold value" and "qualified threshold value" herein refer to any number that is used as a cutoff to characterize a sample such as a test sample containing a nucleic acid from an sm suspected of having a medical condition. The threshold may be compared to a parameter value to determine whether a sample giving rise to such parameter value suggests that the organism has the medical condition. In certain embodiments, a qualified threshold value is calculated using a qualifying data set and serves as a limit of diagnosis of a copy number ion, e.g., an aneuploidy, in an organism. If a threshold is exceeded by results obtained from methods disclosed herein, a subject can be diagnosed with a copy number variation, e.g., trisomy 21. Appropriate threshold values for the methods described herein can be identified by analyzing normalized values (e.g. chromosome doses, NCVs or NSVs) calculated for a training set of s. Threshold values can be identified using qualified (i.e., unaffected) samples in a training set which comprises both ied (i.e., unaffected) s and affected samples. The samples in the training set known to have chromosomal aneuploidies (i.e., the affected samples) can be used to confirm that the chosen thresholds are useful in differentiating affected from unaffected samples in a test set (see the Examples herein). The choice of a threshold is dependent on the level of confidence that the user wishes to have to make the fication. In some embodiments, the training set used to identify riate threshold values comprises at least 10, at least 20, at least , at least 40, at least 50, at least 60, at least 70, at least 80, at least 90, at least 100, at least 200, at least 300, at least 400, at least 500, at least 600, at least 700, at least 800, at least 900, at least 1000, at least 2000 , at least 3000 , at least 4000, or more qualified samples. It may be ageous to use larger sets of qualified samples to improve the diagnostic utility ofthe old values.
The term "bin" refers to a segment of a sequence or a segment of a genome. In some embodiments, bins are contiguous with one another within the genome or some. Each bin may define a sequence of nucleotides in a reference genome. Sizes ofthe bin may be 1 kb, 100 kb, lMb, etc., depending on the analysis required by particular applications and sequence tag density. In addition to their positions within a reference sequence, bins may have other characteristics such as sample coverage and ce structure characteristics such as G-C fraction.
The term "masking threshold" is used herein to refer to a quantity against which a value based on the number of sequence tags in a sequence bin is compared, wherein a bin having a value exceeding the masking threshold is masked.
In some embodiments, the masking threshold can be a tile rank, an absolute count, a mapping quality score, or other suitable values. In some embodiments, a masking threshold may be defined as the percentile rank of a coefficient of ion across multiple unaffected samples. In other embodiments, a masking threshold may be defined as a mapping quality score, e.g., a MapQ score, which relates to the reliability of aligning sequence reads to a reference genome. Note that a masking threshold value is different from a copy number variation (CNV) old value, the latter being a cutoff to characterize a sample containing a nucleic acid from an organism ted of having a medical condition related to CNV. In some embodiment, a CNV threshold value is defined relative to a normalized chromosome value (NCV) or a normalized segment value (NSV) bed elsewhere herein.
] The term lized value" herein refers to a numerical value that s the number of sequence tags identified for the sequence (e.g. chromosome or chromosome t) of interest to the number of sequence tags identified for a normalizing sequence (e.g. normalizing chromosome or normalizing chromosome segment). For example, a "normalized value" can be a chromosome dose as described elsewhere herein, or it can be an NCV, or it can be an NSV as described elsewhere herein.
The term "read" refers to a sequence obtained from a portion of a nucleic acid . Typically, though not necessarily, a read represents a short sequence of contiguous base pairs in the sample. The read may be ented symbolically by the base pair sequence (in A, T, C, or G) of the sample portion. It may be stored in a memory device and processed as appropriate to determine whether it s a reference ce or meets other criteria. A read may be obtained directly from a sequencing tus or indirectly from stored sequence information concerning the sample. In some cases, a read is a DNA sequence of sufficient length (e.g., at least about 25 bp) that can be used to identify a larger sequence or region, e.g., that can be aligned and specifically assigned to a chromosome or genomic region or gene.
The term "genomic read" is used m reference to a read of any segments in the entire genome of an individual.
] The term "sequence tag" is herein used interchangeably with the term I0 "mapped sequence tag" to refer to a sequence read that has been ically assigned, i.e., mapped, to a larger sequence, e.g., a reference genome, by alignment. Mapped sequence tags are uniquely mapped to a reference genome, i.e., they are assigned to a single location to the reference genome. Unless otherwise specified, tags that map to the same ce on a reference sequence are counted once. Tags may be provided as data structures or other assemblages of data. In certain embodiments, a tag contains a read sequence and associated information for that read such as the location of the sequence in the genome, e.g., the position on a chromosome. In certain embodiments, the location is specified for a positive strand orientation. A tag may be d to allow a limited amount of mismatch in aligning to a reference genome. In some embodiments, tags that can be mapped to more than one location on a reference genome, i.e., tags that do not map uniquely, may not be included in the analysis.
The term "non-redundant sequence tag" refers to sequence tags that do not map to the same site, which is counted for the purpose of ining normalized chromosome values (NCVs) in some embodiments. mes multiple sequence reads are aligned to the same locations on a reference genome, yielding redundant or duplicated sequence tags. In some embodiments, duplicate sequence tags that map to the same position are omitted or d as one "non-redundant ce tag" for the e of determining NCVs. In some embodiments, non-redundant sequence tags aligned to cluded sites are counted to yield "non-excluded-site counts" (NES counts) for ining NCVs.
The term "site" refers to a umque position (i.e. chromosome ID, chromosome position and orientation) on a reference genome. In some embodiments, a site may provide a position for a residue, a ce tag, or a segment on a sequence.
"Excluded sites" are sites found in regions of a reference genome that have been excluded for the purpose of counting ce tags. In some embodiments, excluded sites are found in regions of chromosomes that contain repetitive sequences, e.g., centromeres and telomeres, and regions of chromosomes that are common to more than one chromosome, e.g., regions present on the Y- chromosome that are also present on the X some.
"Non-excluded sites" (NESs) are sites that are not excluded m a I0 reference genome for the purpose of counting ce tags. xcluded-site " (NES counts) are the numbers of sequence tags that are mapped to NESs on a reference genome. In some embodiments, NES counts are the numbers of non-redundant sequence tags mapped to NESs. In some ments, coverage and related parameters such normalized ge quantities, global profile removed coverage quantities, and chromosome dose are based on NES counts. In one example, a chromosome dose is calculated as the ratio of the NES count for a chromosome ofinterest to the count for a normalizing chromosome.
Normalized chromosome value (NCV) relates ge of a test sample to coverages of a set of training/qualified samples. In some embodiments, NCV is based on chromosome dose. In some embodiments, NCV relates to the difference between the some dose of a chromosome ofinterest in a test sample and the mean of the corresponding chromosome dose in a set of qualified samples as, and can be calculated as: where P.i and (Ji are the estimated mean and standard deviation, respectively, for the j- th chromosome dose in a set of qualified samples, and xii is the observed j-th chromosome ratio (dose) for test sample i.
In some embodiments, NCV can be calculated "on the fly" by relating the chromosome dose of a chromosome of interest in a test sample to the median of the corresponding chromosome dose in multiplexed samples sequenced on the same flow cells as: X··-M· NCVi1 = i1 ~ J where Mi is the estimated median for the j-th chromosome dose in a set of multiplexed samples ced on the same flow cell; 8i is the standard ion for the j-th chromosome dose in one or more sets of multiplexed samples ced on one or more flow cells, and xii is the observed j-th chromosome dose for test sample i.
In this embodiment, test sample i is one of the multiplexed samples sequenced on the same flow cell from which Mi is determined.
For example, for chromosome of interest 21 in test sample A, which is sequenced as one of 64 multiplexed samples on one flow cell, the NCV for chromosome 21 in test sample A is calculated as the dose of chromosome 21 in sample A minus the median of the dose for chromosome 21 determined in the 64 multiplexed samples, divided by the standard deviation of the dose for chromosome 21 determined for the 64 multiplexed s on flow cell 1, or of additional flow cells.
As used herein, the terms "aligned," "alignment," or "aligning" refer to the process of comparing a read or tag to a reference sequence and thereby determining whether the reference sequence contains the read sequence. If the nce sequence contains the read, the read may be mapped to the reference sequence or, in certain embodiments, to a particular location in the reference sequence. In some cases, alignment simply tells whether or not a read is a member of a particular reference ce (i.e., whether the read is present or absent in the nce sequence). For e, the alignment of a read to the reference sequence for human chromosome 13 will tell whether the read is present in the reference sequence for chromosome 13. A tool that provides this information may be called a set membership tester. In some cases, an alignment additionally tes a location in the reference sequence where the read or tag maps to. For example, ifthe reference sequence is the whole human genome sequence, an alignment may indicate that a read is present on chromosome 13, and may further te that the read is on a particular strand and/or site of chromosome 13. d reads or tags are one or more sequences that are identified as a match in terms ofthe order of their nucleic acid molecules to a known sequence from a reference genome. Alignment can be done manually, although it is typically implemented by a computer thm, as it would be impossible to align reads in a able time period for implementing the methods disclosed herein. One example of an algorithm from aligning sequences is the Efficient Local Alignment of tide Data (ELAND) computer program distributed as part of the Illumina Genomics is pipeline. Alternatively, a Bloom filter or similar set membership tester may be ed to align reads to reference genomes. See US Patent Application No. 61/552,374 filed r 27, 2011 which is incorporated herein by reference in its entirety. The matching of a sequence read in ng can be a 100% sequence match or less than 100% (non-perfect match).
[00127] The term "mapping" used herein refers to specifically ass1gmng a sequence read to a larger sequence, e.g., a reference genome, by alignment.
As used herein, the term "reference genome" or "reference sequence" refers to any particular known genome sequence, whether partial or complete, of any organism or virus which may be used to reference identified sequences from a subject.
For example, a reference genome used for human subjects as well as many other organisms is found at the National Center for Biotechnology Information at lm.nih.gov. A "genome" refers to the complete genetic information of an organism or virus, expressed in nucleic acid sequences.
In various embodiments, the reference sequence is significantly larger than the reads that are aligned to it. For example, it may be at least about 100 times larger, or at least about 1000 times larger, or at least about 10,000 times larger, or at least about 105 times larger, or at least about 106 times larger, or at least about 107 times larger.
In one example, the reference sequence is that of a full length human genome. Such sequences may be referred to as c reference sequences. In r example, the reference sequence is d to a ic human some such as chromosome 13. In some embodiments, a reference Y chromosome is the Y chromosome sequence from human genome version hgl9. Such sequences may be referred to as chromosome reference sequences. Other examples of reference sequences include genomes of other species, as well as chromosomes, subchromosomal regions (such as strands), etc., of any species.
In vanous embodiments, the reference sequence 1s a consensus sequence or other combination derived from multiple individuals. However, in certain applications, the reference sequence may be taken from a ular individual.
The term "clinically-relevant sequence" herein refers to a nucleic acid sequence that is known or is suspected to be ated or implicated with a c or disease condition. ining the absence or presence of a clinically-relevant sequence can be useful in ining a diagnosis or confirming a diagnosis of a medical condition, or providing a sis for the pment of a disease.
[00133] The term "derived" when used in the context of a nucleic acid or a mixture of nucleic acids, herein refers to the means whereby the nucleic acid(s) are obtained from the source from which they originate. For example, in one embodiment, a mixture of nucleic acids that is derived from two different genomes means that the nucleic acids, e.g., cIDNA, were naturally released by cells through naturally occurring ses such as necrosis or apoptosis. In another embodiment, a mixture of nucleic acids that is derived from two different genomes means that the nucleic acids were extracted from two different types of cells from a t.
The term "based on" when used in the context of ing a specific quantitative value, herein refers to using another quantity as input to calculate the specific quantitative value as an output.
The term "patient sample" herein refers to a biological sample obtained from a patient, i.e., a recipient of l attention, care or treatment. The patient sample can be any of the s described herein. In certain embodiments, the patient sample is obtained by non-invasive procedures, e.g., peripheral blood sample or a stool sample. The methods described herein need not be limited to humans.
Thus, various veterinary applications are contemplated in which case the patient sample may be a sample from a non-human mammal (e.g., a feline, a porcine, an equine, a bovine, and the like).
The term "mixed " herein refers to a sample containing a mixture of nucleic acids, which are derived from different genomes.
The term "maternal " herein refers to a biological sample obtained from a pregnant subject, e.g., a woman.
The term "biological fluid" herein refers to a liquid taken from a biological source and includes, for example, blood, serum, plasma, sputum, lavage fluid, ospinal fluid, urine, semen, sweat, tears, saliva, and the like. As used herein, the terms "blood," a" and "serum" expressly ass ons or processed portions thereof. Similarly, where a sample is taken from a biopsy, swab, smear, etc., the "sample" expressly encompasses a processed fraction or portion derived from the biopsy, swab, smear, etc.
] The terms "maternal nucleic acids" and "fetal nucleic acids" herein refer to the nucleic acids of a pregnant female subject and the nucleic acids of the fetus being carried by the pregnant female, respectively.
As used herein, the term "corresponding to" sometimes refers to a nucleic acid sequence, e.g., a gene or a chromosome, that is present in the genome of different subjects, and which does not necessarily have the same sequence in all genomes, but serves to provide the identity rather than the genetic information of a ce ofinterest, e.g., a gene or some.
As used herein, the term "fetal fraction" refers to the fraction of fetal nucleic acids present in a sample sing fetal and maternal nucleic acid. Fetal fraction is often used to characterize the cIDNA in a mother'sblood.
As used herein the term "chromosome" refers to the heredity-bearing gene carrier of a living cell, which is derived from chromatin s comprising DNA and n components (especially hi stones). The conventional internationally recognized dual human genome chromosome numbering system is employed herein.
As used herein, the term "polynucleotide length" refers to the te number of nucleotides in a sequence or in a region of a reference genome. The term "chromosome length" refers to the known length of the chromosome given in base pairs, e.g., provided in the NCBI36/hgl8 assembly of the human chromosome found at lgenomel.lucscl.ledu/cgi-bin/hgTracks?hgsid=l67155613&chromlnfoPage= on the World Wide Web.
[00144] The term "subject" herein refers to a human subject as well as a nonhuman subject such as a mammal, an invertebrate, a vertebrate, a fungus, a yeast, a bacterium, and a virus. Although the examples herein concern humans and the language is primarily directed to human ns, the concepts disclosed herein are applicable to genomes from any plant or animal, and are useful in the fields of veterinary medicine, animal sciences, research laboratories and such.
The term "condition" herein refers to "medical condition" as a broad term that es all es and disorders, but can include injuries and normal health situations, such as pregnancy, that might affect a person's , t from medical assistance, or have implications for l treatments.
The term "complete" when used in reference to a chromosomal aneuploidy herein refers to a gain or loss of an entire chromosome.
[00147] The term "partial" when used in reference to a chromosomal aneuploidy herein refers to a gain or loss of a portion, i.e., segment, of a chromosome.
] The term "mosaic" herein refers to denote the presence of two populations of cells with different karyotypes in one individual who has developed from a single fertilized egg. Mosaicism may result from a mutation during development which is propagated to only a subset ofthe adult cells.
The term "non-mosaic" herein refers to an organism, e.g., a human fetus, ed of cells of one ype.
The term "sensitivity" as used herein refers to the probability that a test result will be positive when the condition ofinterest is present. It may be calculated as the number oftrue positives divided by the sum oftrue positives and false negatives.
The term "specificity" as used herein refers to the probability that a test result will be negative when the condition of interest is absent. It may be calculated as the number oftrue negatives divided by the sum oftrue negatives and false ves.
The term "enrich" herein refers to the process of amplifying polymorphic target nucleic acids contained in a portion of a maternal sample, and combining the amplified product with the remainder of the maternal sample from which the portion was removed. For example, the remainder of the al sample can be the original maternal sample.
The term "original maternal sample" herein refers to a non-enriched ical sample obtained from a pregnant subject, e.g., a woman, who serves as the source from which a portion is removed to amplify polymorphic target nucleic acids.
The "original sample" can be any sample obtained from a pregnant subject, and the processed fractions thereof, e.g., a purified cIDNA sample extracted from a maternal plasma sample.
The term "primer," as used herein refers to an isolated oligonucleotide that is capable of acting as a point of initiation of synthesis when placed under ions inductive to synthesis of an ion product (e.g., the conditions e nucleotides, an inducing agent such as DNA polymerase, and a suitable temperature and pH). The primer is preferably single stranded for maximum efficiency in amplification, but may alternatively be double ed. If double stranded, the primer is first treated to separate its strands before being used to e ion products. Preferably, the primer is an oligodeoxyribonucleotide. The primer must be sufficiently long to prime the synthesis of extension products in the presence of the ng agent. The exact lengths of the s will depend on many factors, including temperature, source of primer, use of the , and the parameters used for primer design.
Introduction and Context CNV in the human genome significantly influence human diversity and predisposition to es (Redon et al., Nature 23 :444-454 [2006], Shaikh et al.
Genome Res 19: 1682-1690 [2009]). Such diseases include, but are not limited to cancer, infectious and autoimmune diseases, diseases of the nervous system, metabolic and/or cardiovascular diseases, and the like.
CNVs have been known to contribute to genetic disease through different mechanisms, resulting in either imbalance of gene dosage or gene disruption in most cases. In addition to their direct correlation with genetic disorders, CNVs are known to mediate phenotypic changes that can be deleterious. Recently, several studies have reported an increased burden of rare or de novo CNVs in complex ers such as Autism, ADHD, and schizophrenia as compared to normal ls, highlighting the potential pathogenicity of rare or unique CNVs (Sebat et al., 316:445 - 449 [2007]; Walsh et al., Science 320:539 - 543 [2008]). CNV arise from genomic rearrangements, primarily owing to deletion, duplication, insertion, and unbalanced ocation events.
It has been shown that cIDNA fragments of fetal origin are shorter, on average, than those of maternal . NIPT nvasive prenatal testing) based on NGS data has been successfully implemented. Current methodologies e sequencing maternal samples using short reads (25bp-36bp), aligning to the genome, computing and normalizing sub-chromosomal coverage, and finally evaluating overrepresentation et chromosomes (13 I 18 I 21 IX I Y) compared to the expected normalized coverage associated with a normal diploid genome. Thus, traditional NIPT assay and analysis relies on the counts or coverage to evaluate the likelihood of fetal aneuploidy.
[00158] Since maternal plasma samples represent a mixture of maternal and fetal cIDNA, the success of any given NIPT method depends on its sensitivity to detect copy number changes in the low fetal on samples. For counting based s, their sensitivity is determined by (a) cing depth and (b) ability of data normalization to reduce technical variance. This disclosure provides ical methodology for NIPT and other applications by deriving fragment size information from, e.g., paired-end reads, and using this information in an analysis pipeline.
Improved analytical sensitivity provides the ability to apply NIPT methods at reduced coverage (e.g., reduced sequencing depth) which enables the use ofthe technology for lower-cost testing of average risk pregnancies.
] Methods, apparatus, and systems are disclosed herein for determining copy number and copy number variations (CNV) of different sequences of interest in a test sample that comprises a mixture of nucleic acids derived from two or more different s, and which are known or are suspected to differ in the amount of one or more sequence ofinterest. Copy number variations determined by the methods and apparatus disclosed herein include gains or losses of entire chromosomes, alterations ing very large chromosomal segments that are microscopically visible, and an abundance of croscopic copy number variation of DNA segments ranging from single nucleotide, to ses (kb), to megabases (Mb) in size.
[00160] In some embodiments, methods are provided for determining copy number variation (CNV) of fetuses using maternal samples containing maternal and fetal cell free DNA Some implementations use nt length (or fragment size) of cIDNA to improve sensitivity and specificity for fetal aneuploidy detection from cIDNA in maternal plasma. Some embodiments are implemented with a PCR free library preparation coupled with paired end DNA sequencing. In some embodiments, both nt size and coverage are utilized to enhance fetal aneuploidy detection. In some embodiments, the methods involve ing independent ng of r fragments with the relative fraction of shorter fragments in bins across the genome.
Some embodiments disclosed herein e methods to improve the sensitivity and/or specificity of ce data analysis by removing within-sample GC-content bias. In some embodiments, removal of within-sample GC-content bias is based on sequence data corrected for systematic variation common across I0 unaffected training samples.
Some embodiments disclosed provide s to derive parameters with high signal to noise ratio from cell free nucleic acid fragments, for determining various genetic conditions related to copy number and CNV, with improved sensitivity, selectivity, and/or efficiency relative to conventional methods. The parameters e, but are not limited to, coverage, fragment size weighted coverage, fraction or ratio of fragments in a defined range, methylation level of fragments, stics obtained from coverage, fetal fraction estimates obtained from coverage ation, etc. The depicted process has been found particularly effective at improving the signal in samples having relatively low fractions of DNA from a genome under eration (e.g., a genome of a fetus). An example of such sample is a maternal blood sample from an individual pregnant with nal twins, ts, etc., where the s assesses copy number variation in the genome of one of the fetuses.
In some embodiments, high analytical sensitivities and specificities can be achieved with a simple library preparation using very low cIDNA input that does not require PCR amplification. The PCR free method simplifies the workflow, improves the tum-around time and eliminates biases that are inherent with PCR methods. In some embodiments, the detection of fetal aneuploidy from maternal plasma can be made more robust and efficient than conventional methods, requiring fewer unique cIDNA fragments. In combination, improved analytical sensitivity and specificity is achieved with a very fast turnaround time at a significantly lower number of cIDNA fragments. This potentially allows NIPT to be carried out at significantly lower costs to facilitate application in the general obstetric population.
In various implementations, PCR-free library preparation is possible with the disclosed methods. Some implementations eliminate nt biases of PCR methods, d assay complexity, reduce required sequencing depth (2.5X lower), provide faster turnaround time, e.g., turn around in one day, enable in-process fetal fraction (FF) measurement, facilitate discrimination between maternal and fetal/placental cIDNA using fragment size information.
Evaluating CNV Methods for determination ofCNV Using the sequence coverage value, fragment size parameters, and/or I0 ation levels provided by the methods disclosed herein, one can determine various genetic conditions related to copy number and CNV of ces, chromosomes, or chromosome segments with improved ivity, selectivity, and/or efficiency ve to using sequence ge values obtained by conventional methods. For example, in some embodiments, the masked reference sequences are used for determining the presence or absence of any two or more ent complete fetal chromosomal aneuploidies in a maternal test sample comprising fetal and maternal nucleic acid molecules. Exemplary methods provided below align reads to reference sequences (including reference genomes). The alignment can be performed on an unmasked or masked reference ce, thereby yielding ce tags mapped to the reference sequence. In some ments, only sequence tags falling on unmasked segments of the reference sequence are taken into account to determine copy number variation.
In some embodiments, mg a nucleic acid sample for CNV involves characterizing the status of a chromosomal or segment aneuploidy by one of three types of calls: "normal" or "unaffected," "affected," and "no-call." Thresholds for calling normal and affected are typically set. A parameter related to aneuploidy or other copy number variation is ed in a sample and the measured value is compared to the thresholds. For duplication type aneuploidies, a call of affected is made if a chromosome or segment dose (or other measured value sequence content) is above a d threshold set for affected samples. For such aneuploidies, a call of normal is made if the chromosome or segment dose is below a threshold set for normal s. By contrast for deletion type aneuploidies, a call of affected is made 2016/067886 if a chromosome or segment dose is below a d threshold for affected samples, and a call of normal is made if the chromosome or t dose is above a threshold set for normal samples. For example, in the presence of trisomy the "normal" call is determined by the value of a parameter, e.g., a test chromosome dose that is below a user-defined threshold of reliability, and the "affected" call is determined by a parameter, e.g., a test chromosome dose, that is above a user-defined threshold of reliability. A "no-call" result is determined by a parameter, e.g., a test chromosome dose that lies n the thresholds for making a "normal" or an "affected" call. The term "no-call" is used interchangeably with "unclassified".
[00167] The parameters that may be used to determine CNV include, but are not limited to, coverage, fragment size biased/weighted ge, fraction or ratio of nts in a defined size range, and methylation level of fragments. As discussed herein, coverage is obtained from counts of reads aligned to a region of a reference genome and optionally normalized to produce sequence tag counts. In some embodiments, ce tag counts can be weighted by nt size.
In some embodiments, a fragment size parameter is biased toward fragment sizes characteristic of one of the genomes. A fragment size parameter is a parameter that relates to the size of a fragment. A parameter is biased toward a fragment size when: 1) the ter is favorably weighted for the fragment size, e.g., a count ed more heavily for the size than for other sizes; or 2) the parameter is obtained from a value that is favorably weighted for the fragment size, e.g., a ratio obtained from a count weighted more heavily for the size. A size is characteristic of a genome when the genome has an enriched or higher concentration of nucleic acid of the size relative to another genome or another portion ofthe same genome.
] In some embodiments, the method for determining the presence or absence of any complete fetal chromosomal aneuploidies in a maternal test sample comprises (a) ing sequence information for fetal and maternal nucleic acids in the maternal test sample; (b) using the sequence information and the method described above to identify a number of sequence tags, sequence coverage quantity, a fragment size parameter, or another parameter for each of the chromosomes of interest selected from chromosomes 1-22, X and Y and to identify a number of sequence tags or another parameter for one or more normalizing chromosome sequences; (c) using the number of sequence tags or the other parameter identified for 2016/067886 each of the chromosomes of interest and the number of sequence tags or the other parameter identified for each of the normalizing chromosomes to calculate a single chromosome dose for each of the chromosomes of interests; and (d) comparing each some dose to a threshold value, and thereby determining the presence or absence of any complete fetal chromosomal aneuploidies in the maternal test sample.
In some embodiments, step (a) described above can comprise sequencing at least a portion of the nucleic acid molecules of a test sample to obtain said sequence information for the fetal and maternal nucleic acid molecules ofthe test sample. In some embodiments, step (c) comprises ating a single chromosome I0 dose for each of the chromosomes of interest as the ratio of the number of sequence tags or the other parameter identified for each of the somes of interest and the number of sequence tags or the other parameter identified for the normalizing chromosome sequence(s). In some other embodiments, chromosome dose is based on processed sequence coverage ties derived from the number of sequence tags or another parameter. In some embodiments, only unique, dundant sequence tags are used to calculate the processed ce coverage quantities or another parameter.
In some embodiments, the processed sequence ge quantity is a sequence tag density ratio, which is the number of sequence tag standardized by sequence length.
In some embodiments, the sed sequence coverage quantity or the other ter is a normalized sequence tag or another normalized parameter, which is the number of sequence tags or the other parameter of a sequence of interest divided by that of all or a substantial portion of the genome. In some embodiments, the processed sequence coverage quantity or the other parameter such as a fragment size parameter is adjusted according to a global profile of the sequence of st. In some ments, the processed sequence coverage quantity or the other parameter is ed according to the within-sample correlation between the GC content and the sequence coverage for the sample being tested. In some embodiments, the processed sequence coverage quantity or the other parameter results from combinations of these processes, which are r described elsewhere herein.
] In some embodiments, a chromosome dose is calculated as the ratio of the processed sequence coverage or the other parameter for each of the chromosomes rest and that for the normalizing chromosome sequence(s). 2016/067886 In any one of the embodiments above, the complete chromosomal oidies are ed from complete chromosomal trisomies, complete chromosomal monosomies and complete chromosomal polysomies. The complete chromosomal aneuploidies are selected from complete oidies of any one of chromosome 1-22, X, and Y. For example, the said ent complete fetal chromosomal aneuploidies are selected from trisomy 2, trisomy 8, trisomy 9, trisomy , trisomy 21, trisomy 13, trisomy 16, trisomy 18, trisomy 22, , 47,XXX, 47,XYY, and monosomy X.
In any one ofthe embodiments above, steps (a)-(d) are repeated for test samples from different maternal subjects, and the method comprises ining the presence or absence of any two or more different te fetal chromosomal oidies in each ofthe test s.
In any one ofthe embodiments above, the method can further comprise calculating a normalized chromosome value (NCV), wherein the NCV relates the chromosome dose to the mean of the corresponding chromosome dose in a set of qualified samples as: where µ1 and 81 are the estimated mean and standard deviation, respectively, for the jth chromosome dose in a set of qualified samples, and xii is the observed j-th some dose for test sample i.
[00175] In some embodiments, NCV can be calculated "on the fly" by relating the chromosome dose of a chromosome of interest in a test sample to the median of the corresponding chromosome dose in multiplexed samples sequenced on the same flow cells as: X··-M· NCV,. · = iJ 1 l] (Ji~ where Mi is the estimated median for the j-th chromosome dose in a set of multiplexed samples sequenced on the same flow cell; 8i is the standard deviation for the j-th chromosome dose in one or more sets of multiplexed samples sequenced on one or more flow cells, and xi is the observed j-th chromosome dose for test sample i.
In this embodiment, test sample i is one of the multiplexed samples ced on the same flow cell from which Mi is determined.
In some embodiments, a method 1s provided for determining the presence or absence of different partial fetal chromosomal aneuploidies in a maternal test sample comprising fetal and maternal nucleic acids. The method involves procedures analogous to the method for detecting te aneuploidy as outlined above. However, instead of analyzing a complete some, a segment of a chromosome is analyzed. See US Patent Application Publication No. 2013/0029852, which is incorporated by reference.
[00177] Figure 1 shows a method for determining the presence of copy number variation in accordance with some embodiments. s 100 illustrated in Figure 1 uses sequence tag ge based on the number of sequence tags (i.e., the sequence tag count) to determine CNV. However, similar to the description above for calculation of a NCV, other variables or ters, such as size, size ratio, and methylation level, may be used instead of coverage. In some implementations, two or more variables are ed to ine a CNV. Furthermore, coverage and other parameters may be weighted based on the size of the fragments from which tags are derived. For ease of reading, only coverage is referred to in s 100 illustrated in Figure 1, but one should note that other parameters, such as size, size ratio, and methylation level, count weighted by size, etc. may be used in place of coverage.
In operations 130 and 135, qualified sequence tag coverages (or values of another parameter) and test sequence tag coverages (or values of another parameter) are determined. The present sure provides processes to determine coverage quantities that provide improved sensitivity and selectivity relative to conventional methods. Operation 130 and 135 are marked by sks and emphasized by boxes of heavy lines to indicate these operations bute to improvement over prior art. In some embodiments, the sequence tag coverage quantities are normalized, adjusted, trimmed, and otherwise processed to improve the sensitivity and selectivity of the analysis. These processes are further described elsewhere herein.
From an over-view perspective, the method makes use of normalizing sequences of qualified training samples in ination of CNV of test samples. In some ments, the qualified training s are unaffected and have normal copy number. Normalizing sequences provide a mechanism to normalize ements for intra-run and inter-run variabilities. Normalizing sequences are identified using sequence information from a set of qualified samples obtained from subjects known to comprise cells having a normal copy number for any one sequence of interest, e.g., a chromosome or segment thereof. Determination of normalizing sequences is outlined in steps 110, 120, 130, 145 and 146 of the embodiment of the method depicted in Figure 1. In some embodiments, the normalizing sequences are used to calculate sequence dose for test sequences. See step 150. In some embodiments, normalizing sequences are also used to calculate a threshold against which the sequence dose of the test sequences is compared. See step 150. The ce information obtained from the izing sequence and the test sequence is used for determining statistically meaningful identification of chromosomal oidies in test samples (step 160).
[00180] Turning to the details of the method for determining the presence of copy number variation according to some ments, Figure 1 provides a flow diagram 100 of an embodiment for ining a CNV of a sequence of interest, e.g., a chromosome or segment f, in a biological sample. In some ments, a biological sample is obtained from a subject and ses a mixture of nucleic acids contributed by different genomes. The different genomes can be contributed to the sample by two duals, e.g., the ent genomes are contributed by the fetus and the mother carrying the fetus. Also, the different genomes can be contributed to the sample by three or more individuals, e.g., the different genomes are contributed by two or more fetuses and the mother carrying the fetuses. Alternatively, the genomes are contributed to the sample by aneuploid cancerous cells and normal euploid cells from the same subject, e.g., a plasma sample from a cancer patient.
Apart from analyzing a patient's test sample, one or more normalizing chromosomes or one or more normalizing chromosome segments are selected for each possible chromosome of interest. The normalizing chromosomes or segments are identified asynchronously from the normal testing of patient samples, which may take place in a clinical setting. In other words, the normalizing chromosomes or segments are fied prior to testing patient samples. The associations between normalizing chromosomes or ts and chromosomes or segments of interest are stored for use during testing. As explained below, such association is typically maintained over periods of time that span testing of many samples. The following discussion concerns embodiments for selecting normalizing chromosomes or some segments for individual chromosomes or segments ofinterest.
[00182] A set of qualified s is obtained to identify qualified normalizing ces and to provide variance values for use in determining statistically meaningful identification of CNV in test samples. In step 110, a plurality of biological qualified samples are obtained from a plurality of subjects known to comprise cells having a normal copy number for any one sequence of interest. In one embodiment, the qualified samples are obtained from mothers pregnant with a fetus that has been confirmed using cytogenetic means to have a normal copy number of chromosomes. The biological ied samples may be a biological fluid, e.g., plasma, or any suitable sample as described below. In some embodiments, a qualified sample contains a mixture ofnucleic acid molecules, e.g., cIDNA les. In some embodiments, the qualified sample is a maternal plasma sample that contains a mixture of fetal and maternal cIDNA molecules. Sequence information for normalizing chromosomes and/or ts thereof is obtained by sequencing at least a portion of the nucleic acids, e.g., fetal and maternal nucleic acids, using any known sequencing method. Preferably, any one of the Next Generation Sequencing (NGS) s described elsewhere herein is used to sequence the fetal and maternal nucleic acids as single or clonally amplified molecules. In s embodiments, the qualified samples are processed as disclosed below prior to and during sequencing.
They may be processed using apparatus, systems, and kits as disclosed .
] In step 120, at least a portion of each of all the qualified nucleic acids contained in the qualified samples are sequenced to generate millions of sequence reads, e.g., 36bp reads, which are aligned to a reference genome, e.g., hgl8. In some embodiments, the ce reads comprise about 20bp, about 25bp, about 30bp, about35bp,about40bp,about45bp, about50bp,about55bp,about60bp, about65bp, about 70bp, about 75bp, about 80bp, about 85bp, about90bp, about 95bp, about lOObp,about llObp,about 120bp,about 130,about about 150bp,about200bp, about 250bp, about 300bp, about 350bp, about 400bp, about 450bp, or about 500bp.
It is expected that technological es will enable single-end reads of greater than 500bp enabling for reads of greater than about 1OOObp when paired end reads are generated. In one embodiment, the mapped sequence reads se 36bp. In another embodiment, the mapped sequence reads comprise 25bp.
Sequence reads are aligned to a reference genome, and the reads that are uniquely mapped to the reference genome are known as sequence tags. Sequence tags falling on masked segments of a masked reference sequence are not counted for is of CNV.
In one embodiment, at least about 3 x 106 qualified sequence tags, at least about 5 x 106 qualified sequence tags, at least about 8 x 106 qualified sequence tags, at least about 10 x 106 qualified sequence tags, at least about 15 x 106 qualified sequence tags, at least about 20 x 106 qualified sequence tags, at least about 30 x 106 qualified sequence tags, at least about 40 x 106 qualified ce tags, or at least about 50 x 106 qualified sequence tags comprising between 20 and 40bp reads are ed from reads that map uniquely to a nce genome.
In step 130, all the tags obtained from cing the nucleic acids in the qualified samples are counted to obtain a qualified sequence tag ge.
Similarly, in operation 135, all tags obtained from a test sample are counted to obtain a test sequence tag coverage. The present disclosure provides ses to determine coverage ties that provides improved sensitivity and selectivity relative to conventional methods. Operation 130 and 135 are marked by asterisks and emphasized by boxes of heavy lines to indicate these operations contribute to improvement over prior art. In some embodiments, the sequence tag coverage ties are normalized, ed, trimmed, and otherwise processed to improve the sensitivity and selectivity of the analysis. These processes are further described elsewhere herein.
[00187] As all qualified sequence tags are mapped and counted in each of the qualified samples, the sequence tag coverage for a sequence of interest, e.g., a clinically-relevant ce, in the qualified samples is determined, as are the sequence tag coverages for additional sequences from which normalizing sequences are identified subsequently.
[00188] In some embodiments, the sequence of interest is a chromosome that is associated with a complete chromosomal aneuploidy, e.g., chromosome 21, and the qualified normalizing sequence is a complete chromosome that is not associated with a somal aneuploidy and whose variation m ce tag coverage approximates that of the sequence (i.e., chromosome) of interest, e.g., chromosome 21. The selected normalizing chromosome(s) may be the one or group that best approximates the variation in sequence tag coverage of the sequence of interest. Any one or more of chromosomes 1-22, X, and Y can be a sequence ofinterest, and one or more chromosomes can be identified as the normalizing sequence for each of the any one chromosomes 1-22, X and Y in the qualified samples. The normalizing chromosome can be an individual chromosome or it can be a group of chromosomes as described elsewhere herein.
[00189] In another embodiment, the sequence of interest is a segment of a some associated with a partial aneuploidy, e.g., a chromosomal deletion or ion, or unbalanced chromosomal translocation, and the normalizing sequence is a somal segment (or group of segments) that is not associated with the partial aneuploidy and whose variation in sequence tag coverage approximates that of the chromosome segment associated with the partial aneuploidy. The selected normalizing chromosome segment(s) may be the one or more that best approximates the variation in sequence tag coverage of the sequence of interest. Any one or more segments of any one or more chromosomes 1-22, X, and Y can be a sequence of interest.
[00190] In other ments, the sequence of interest is a segment of a some associated with a partial aneuploidy and the normalizing sequence is a whole chromosome or chromosomes. In still other embodiments, the sequence of interest is a whole some associated with an oidy and the izing sequence 1s a chromosomal segment or segments that are not associated with the aneuploidy.
Whether a single sequence or a group of sequences are identified in the qualified samples as the izing sequence(s) for any one or more sequences of interest, the qualified normalizing sequence may be chosen to have a variation in sequence tag coverage or a fragment size parameter that best or effectively approximates that of the ce of interest as determined in the qualified samples.
For example, a ied normalizing sequence is a sequence that produces the smallest variability across the qualified samples when used to normalize the sequence of interest, i.e., the variability of the normalizing sequence is closest to that of the sequence of interest determined in qualified s. Stated another way, the qualified izing sequence is the sequence selected to produce the least variation in sequence dose (for the sequence ofinterest) across the qualified samples. Thus, the process selects a sequence that when used as a normalizing chromosome is expected to e the smallest variability in run-to-run chromosome dose for the sequence of interest.
The normalizing sequence fied in the qualified samples for any one or more sequences of st remains the normalizing sequence of choice for determining the presence or e of aneuploidy in test samples over days, weeks, I0 months, and possibly years, provided that procedures needed to generate sequencing libraries, and sequencing the samples are essentially unaltered over time. As described above, normalizing sequences for determining the presence of aneuploidies are chosen for (possibly among other reasons as well) the variability in the number of ce tags or values of the nt size parameter that are mapped to it among samples, e.g., ent samples, and sequencing runs, e.g., cing runs that occur on the same day and/or different days, that best approximates the variability of the sequence of interest for which it is used as a normalizing parameter. Substantial alterations in these procedures will affect the number of tags that are mapped to all sequences, which in turn will determine which one or group of sequences will have a variability across samples in the same and/or in different sequencing runs, on the same day or on different days that most closely approximates that of the sequence(s) of interest, which would e that the set of normalizing sequences be redetermined.
Substantial alterations in procedures include changes in the laboratory protocol used for preparing the sequencing library, which includes changes related to preparing samples for multiplex cing instead of singleplex sequencing, and changes in sequencing rms, which include changes in the chemistry used for sequencmg.
In some embodiments, the normalizing sequence chosen to normalize a particular sequence of interest is a sequence that best distinguishes one or more qualified, samples from one or more affected samples, which implies that the normalizing sequence is a sequence that has the greatest differentiability, i.e., the differentiability of the normalizing sequence is such that it provides optimal differentiation to a ce of interest in an affected test sample to easily distinguish the affected test sample from other unaffected samples. In other embodiments, the izing sequence is a sequence that has a ation of the smallest variability and the greatest differentiability.
The level of differentiability can be determined as a statistical difference between the sequence doses, e.g., chromosome doses or segment doses, in a population of qualified samples and the chromosome dose(s) in one or more test samples as described below and shown in the Examples. For example, differentiability can be represented numerically as a t-test value, which ents the statistical difference between the chromosome doses in a population of qualified samples and the chromosome dose(s) in one or more test samples. Similarly, differentiability can be based on t doses instead of chromosome doses.
Alternatively, differentiability can be represented numerically as a Normalized Chromosome Value (NCV), which is a z-score for chromosome doses as long as the bution for the NCV is normal. rly, in the case where chromosome segments are the sequences of interest, differentiability of segment doses can be ented numerically as a Normalized Segment Value (NSV), which is a z-score for chromosome segment doses as long as the distribution for the NSV is normal. In determining the z-score, the mean and standard deviation of chromosome or segment doses in a set of qualified samples can be used. Alternatively, the mean and rd deviation of chromosome or segment doses in a training set sing qualified samples and affected samples can be used. In other embodiments, the normalizing sequence 1s a sequence that has the st variability and the greatest differentiability or an optimal combination of small variability and large differentiability.
[00195] The method identifies sequences that inherently have r characteristics and that are prone to similar variations among samples and sequencing runs, and which are useful for determining sequence doses in test samples.
Determination ofsequence doses In some ments, chromosome or t doses for one or more chromosomes or segments of interest are determined in all qualified samples as described in step 146 shown in Figure 1, and a normalizing chromosome or segment sequence is identified in step 145. Some izing sequences are provided before sequence doses are calculated. Then one or more izing sequences are identified according to various criteria as further described below, see step 145. In some embodiments, e.g., the identified normalizing sequence results in the st variability in sequence dose for the sequence rest across all qualified samples.
[00197] In step 146, based on the calculated qualified tag densities, a qualified sequence dose, i.e., a chromosome dose or a t dose, for a sequence of interest is determined as the ratio of the sequence tag coverage for the sequence of st and the qualified ce tag coverage for additional sequences from which normalizing sequences are identified subsequently in step 145. The identified normalizing sequences are used subsequently to determine sequence doses in test samples.
In one embodiment, the sequence dose in the qualified samples is a chromosome dose that is calculated as the ratio of the number of sequence tags or fragment size parameter for a chromosome of interest and the number of sequence tags for a normalizing chromosome sequence in a qualified sample. The normalizing chromosome sequence can be a single some, a group of chromosomes, a t of one chromosome, or a group of segments from ent chromosomes.
Accordingly, a chromosome dose for a chromosome of interest is determined in a qualified sample as the ratio of the number of tags for a chromosome of interest and the number of tags for (i) a normalizing chromosome sequence composed of a single chromosome, (ii) a normalizing chromosome sequence composed of two or more chromosomes, (iii) a normalizing segment sequence composed of a single segment of a chromosome, (iv) a normalizing t sequence composed of two or more segments form one chromosome, or (v) a izing t sequence composed of two or more segments of two or more chromosomes. Examples for determining a chromosome dose for chromosome of interest 21 according to (i)-(v) are as follows: chromosome doses for chromosome of interest, e.g., chromosome 21, are determined as a ratio of the ce tag coverage of some 21 and one of the following sequence tag coverages: (i) each of all the remaining chromosomes, i.e., chromosomes 1-20, chromosome 22, chromosome X, and some Y; (ii) all possible combinations of two or more remaining chromosomes; (iii) a segment of another chromosome, e.g., chromosome 9; (iv) two segments of one other chromosome, e.g., two segments of chromosome 9; (v) two ts oftwo different chromosomes, e.g., a segment of chromosome 9 and a segment of chromosome 14.
In r ment, the sequence dose in the qualified samples is a segment dose as d to a chromosome dose, which segment dose is calculated as the ratio of the number of sequence tags for a segment of interest, that is not a whole some, and the number of sequence tags for a normalizing segment sequence in a qualified sample. The normalizing segment sequence can be any ofthe normalizing some or segment sequences discussed above.
Identification ofnormalizing sequences ] In step 145, a normalizing sequence is identified for a sequence of interest. In some embodiments, e.g., the normalizing sequence is the sequence based on the calculated sequence doses, e.g., that result in the smallest variability in sequence dose for the ce of interest across all qualified training samples. The method identifies sequences that inherently have similar characteristics and are prone to similar variations among samples and sequencing runs, and which are useful for determining sequence doses in test samples.
Normalizing sequences for one or more sequences of st can be fied in a set of qualified samples, and the sequences that are identified in the qualified samples are used subsequently to calculate sequence doses for one or more sequences of interest in each of the test samples (step 150) to determine the presence or e of aneuploidy in each of the test samples. The normalizing sequence identified for chromosomes or segments of interest may differ when different sequencing platforms are used and/or when ences exist in the purification of the c acid that is to be sequenced and/or preparation of the sequencing library. The use of izing sequences according to the methods described herein provides specific and sensitive measure of a variation in copy number of a chromosome or segment thereof irrespective of sample preparation and/or sequencing platform that is used.
In some embodiments, more than one normalizing sequence is fied, i.e., different normalizing sequences can be determined for one sequence of interest, and multiple sequence doses can be determined for one sequence of interest.
For example, the variation, e.g., coefficient of variation (CV= standard deviation/mean), in chromosome dose for chromosome ofinterest 21 is least when the ce tag coverage of chromosome 14 is used. However, two, three, four, five, six, seven, eight or more normalizing sequences can be identified for use m determining a sequence dose for a sequence of interest in a test sample. As an example, a second dose for chromosome 21 in any one test sample can be determined using chromosome 7, chromosome 9, chromosome 11 or chromosome 12 as the normalizing chromosome sequence as these chromosomes all have CV close to that for chromosome 14.
In some ments, when a single chromosome is chosen as the izing chromosome sequence for a chromosome of interest, the izing chromosome sequence will be a chromosome that results in chromosome doses for the chromosome of interest that has the smallest variability across all samples tested, e.g., qualified samples. In some instances, the best izing chromosome may not have the least ion, but may have a distribution of ied doses that best distinguishes a test sample or samples from the qualified s, i.e., the best normalizing chromosome may not have the lowest variation, but may have the greatest differentiability.
In some embodiments, normalizing sequences include one or more robust autosomes sequences or segments f. In some embodiments, the robust autosomes include all autosomes except for the chromosome(s) of interest. In some embodiments, the robust autosomes include all autosomes except for chr X, Y, 13, 18, and 21. In some embodiments, the robust autosomes include all autosomes except those determined from a sample to be deviating from a normal diploid state, which can be useful in determining cancer genomes that have abnormal copy number relative to a normal diploid genome.
Determination ofaneuploidies in test samples Based on the fication of the normalizing sequence(s) in qualified s, a sequence dose is determined for a sequence of interest in a test sample comprising a mixture of nucleic acids derived from genomes that differ in one or more sequences of interest.
In step 115, a test sample is obtained from a subject suspected or known to carry a clinically-relevant CNV of a sequence of interest. The test sample may be a biological fluid, e.g., plasma, or any suitable sample as described below. As explained, the sample may be obtained using a non-invasive procedure such as a simple blood draw. In some embodiments, a test sample contains a mixture ofnucleic acid molecules, e.g., cIDNA molecules. In some embodiments, the test sample is a maternal plasma sample that ns a mixture of fetal and maternal cIDNA molecules.
] In step 125, at least a portion ofthe test nucleic acids in the test sample is sequenced as described for the ied samples to te millions of sequence reads, e.g., 36bp reads. In various embodiments, 2x36 bp paired end reads are used for paired end sequencing. As in step 120, the reads generated from cing the nucleic acids in the test sample are uniquely mapped or aligned to a reference genome to produce tags. As described in step 120, at least about 3 x 106 qualified sequence tags, at least about 5 x 106 qualified sequence tags, at least about 8 x 106 qualified ce tags, at least about 10 x 106 qualified sequence tags, at least about 15 x 106 qualified sequence tags, at least about 20 x 106 qualified sequence tags, at least about x 106 ied ce tags, at least about 40 x 106 qualified sequence tags, or at least about 50 x 106 qualified sequence tags comprising between 20 and 40bp reads are obtained from reads that map uniquely to a reference genome. In certain embodiments, the reads produced by sequencing apparatus are provided in an electronic format. Alignment is accomplished using computational apparatus as discussed below. Individual reads are compared against the reference genome, which is often vast (millions of base pairs) to fy sites where the reads uniquely correspond with the reference genome. In some embodiments, the alignment ure permits limited mismatch between reads and the reference . In some cases, 1, 2, or 3 base pairs in a read are permitted to mismatch ponding base pairs in a reference genome, and yet a mapping is still made.
In step 135, all or most of the tags obtained from sequencing the nucleic acids in the test samples are counted to determine a test sequence tag coverage using a computational apparatus as described below. In some embodiments, each read is aligned to a particular region of the reference genome (a some or segment in most cases), and the read is converted to a tag by appending site information to the read. As this process unfolds, the computational apparatus may keep a running count of the number of tags/reads mapping to each region of the WO 36059 reference genome osome or segment in most cases). The counts are stored for each chromosome or segment of interest and each corresponding normalizing chromosome or t.
In certain embodiments, the reference genome has one or more excluded regions that are part of a true biological genome but are not ed in the reference genome. Reads potentially aligning to these excluded regions are not counted. Examples of excluded regions include regions of long repeated sequences, s of similarity between X and Y chromosomes, etc. Using a masked reference sequence obtained by masking techniques described above, only tags on unmasked segments ofthe reference sequence are taken into account for analysis of CNV.
In some embodiments, the method determines whether to count a tag more than once when multiple reads align to the same site on a reference genome or ce. There may be occasions when two tags have the same sequence and therefore align to an identical site on a reference ce. The method employed to count tags may under certain stances exclude from the count identical tags deriving from the same sequenced sample. If a disproportionate number of tags are identical in a given sample, it suggests that there is a strong bias or other defect in the procedure. ore, in accordance with certain embodiments, the counting method does not count tags from a given sample that are identical to tags from the sample that were previously counted.
Various criteria may be set for choosing when to disregard an identical tag from a single sample. In certain ments, a defined percentage of the tags that are counted must be unique. Ifmore tags than this threshold are not unique, they are disregarded. For example, if the defined percentage requires that at least 50% are unique, identical tags are not counted until the percentage ofunique tags s 50% for the sample. In other embodiments, the threshold number ofunique tags is at least about 60%. In other embodiments, the threshold percentage of unique tags is at least about 75%, or at least about 90%, or at least about 95%, or at least about 98%, or at least about 99%. A threshold may be set at 90% for chromosome 21. If30M tags are aligned to chromosome 21, then at least 27M of them must be unique. If3M counted tags are not unique and the 30 million and first tag is not unique, it is not counted.
The choice ofthe particular threshold or other criterion used to ine when not to count further identical tags can be selected using riate statistical analysis. One factor influencing this threshold or other criterion is the relative amount of sequenced sample to the size of the genome to which tags can be aligned. Other factors include the size ofthe reads and similar considerations.
In one embodiment, the number of test sequence tags mapped to a ce of interest is normalized to the known length of a sequence of interest to which they are mapped to provide a test sequence tag density ratio. As described for the qualified samples, normalization to the known length of a sequence of interest is not required, and may be included as a step to reduce the number of digits in a number to simplify it for human interpretation. As all the mapped test sequence tags are counted in the test sample, the sequence tag coverage for a sequence ofinterest, e.g., a clinically-relevant sequence, in the test samples is determined, as are the sequence tag coverages for additional sequences that correspond to at least one izing sequence identified in the qualified samples.
In step 150, based on the identity of at least one normalizing sequence in the qualified samples, a test sequence dose is determined for a sequence of interest in the test sample. In various ments, the test sequence dose is computationally determined using the sequence tag coverages of the sequence of interest and the corresponding normalizing sequence as described . The computational apparatus responsible for this undertaking will onically access the association between the sequence of interest and its associated normalizing sequence, which may be stored in a database, table, graph, or be ed as code in m instructions.
] As bed elsewhere herein, the at least one normalizing sequence can be a single sequence or a group of sequences. The ce dose for a sequence of interest in a test sample is a ratio of the sequence tag coverage determined for the sequence of st in the test sample and the sequence tag coverage of at least one normalizing sequence determined in the test sample, wherein the normalizing sequence in the test sample ponds to the normalizing sequence identified in the qualified samples for the particular sequence of interest. For example, if the normalizing sequence identified for chromosome 21 in the qualified samples is determined to be a some, e.g., chromosome 14, then the test sequence dose for chromosome 21 nce of st) is determined as the ratio of the sequence tag coverage for chromosome 21 in and the sequence tag coverage for chromosome 14 each determined in the test sample. Similarly, chromosome doses for chromosomes 13, 18, X, Y, and other chromosomes associated with chromosomal oidies are determined. A normalizing sequence for a chromosome of interest can be one or a group of chromosomes, or one or a group of chromosome segments. As described previously, a sequence of interest can be part of a chromosome, e.g., a chromosome segment. Accordingly, the dose for a chromosome segment can be determined as the ratio of the sequence tag coverage determined for the segment in the test sample and the ce tag coverage for the normalizing chromosome segment in the test sample, wherein the normalizing segment in the test sample corresponds to the normalizing segment (single or a group of segments) identified in the ied samples for the particular segment of interest. Chromosome segments can range from kilobases (kb) to megabases (Mb) in size (e.g., about lkb to 10 kb, or about 10 kb to 100 kb, or about lOOkb to 1 Mb).
In step 155, threshold values are derived from standard deviation values established for qualified ce doses determined in a plurality of qualified samples and sequence doses determined for samples known to be aneuploid for a sequence of interest. Note that this operation is typically performed asynchronously with analysis of patient test samples. It may be performed, for example, concurrently with the selection of izing sequences from qualified samples. Accurate classification depends on the differences between probability distributions for the different s, i.e., type of aneuploidy. In some examples, thresholds are chosen from empirical distribution for each type of aneuploidy, e.g., trisomy 21. Possible threshold values that were ished for classifying trisomy 13, y 18, trisomy 21, and my X aneuploidies as described in the Examples, which describe the use of the method for determining chromosomal aneuploidies by sequencing cIDNA ted from a maternal sample comprising a mixture of fetal and maternal nucleic acids. The threshold value that is determined to guish samples affected for an aneuploidy of a chromosome can be the same or can be different from the threshold for a different aneuploidy. As is shown in the Examples, the threshold value for each chromosome of interest is determined from the variability in the dose of the chromosome of interest across samples and sequencing runs. The less variable the chromosome dose for any chromosome ofinterest, the narrower the spread in the dose for the some of st across all the cted s, which are used to set the threshold for determining different aneuploidies.
Returning to the process flow associated with classifying a patient test sample, m step 160, the copy number variation of the sequence of interest is determined in the test sample by comparing the test sequence dose for the sequence of interest to at least one threshold value ished from the qualified sequence doses.
This operation may be performed by the same computational apparatus employed to measure sequence tag coverages and/or calculate segment doses.
In step 160, the calculated dose for a test ce of interest is compared to that set as the threshold values that are chosen according to a userdefined "threshold of reliability" to fy the sample as a "normal" an "affected" or a "no call." The "no call" samples are samples for which a definitive diagnosis cannot be made with reliability. Each type of affected sample (e.g., trisomy 21, l trisomy 21, monosomy X) has its own thresholds, one for calling normal (unaffected) samples and another for calling affected samples (although in some cases the two thresholds coincide). As described elsewhere herein, under some stances a no-call can be converted to a call ted or normal) if fetal fraction of nucleic acid in the test sample is sufficiently high. The fication of the test sequence may be reported by the computational apparatus employed in other operations of this process flow. In some cases, the classification is reported in an electronic format and may be displayed, d, , etc. to interest s.
[00218] In some embodiments, the determination of CNV comprises calculating a NCV or NSV that relates the some or segment dose to the mean of the corresponding chromosome or segment dose in a set of qualified samples as described above. Then CNV can be determined by comparing the NCV/NSV to a predetermined copy number evaluation threshold value.
[00219] The copy number evaluation threshold can be chosen to optimize the rate of false positives and false negatives. The higher the copy number evaluation threshold, the less likely the occurrence of a false positive. Similarly, the lower the threshold, the less likely the occurrence of a false negative. Thus, a trade-off exists between a first ideal threshold above which only true positives are classified, and a second ideal threshold below which only true negatives are classified.
Thresholds are set largely depending on the variability in chromosome doses for a particular chromosome of st as determined in a set of unaffected 2016/067886 samples. The variability is dependent on a number of s, including the fraction offetal cDNA present in a sample. The variability (CV) is determined by the mean or median and standard deviation for chromosome doses across a population of unaffected samples. Thus, the threshold (s) for classifying aneuploidy use NCVs, according to : (where µ1 and 81 are the estimated mean and rd ion, respectively, for thej-th chromosome dose in a set of qualified samples, and xi1 is the observedj-th chromosome dose for test sample i.) with an associated fetal fraction as: FFij = 2 X INC~x&jl = 2 X NCV X CV Thus, for every NCV of a chromosome of interest, an expected fetal fraction associated with the given NCV value can be calculated from the CV based on the mean and standard deviation of the chromosome ratio for the some of interest across a population fected samples.
Subsequently, based on the relationship between fetal fraction and NCV values, a decision ry can be chosen above which samples are determined to be positive (affected) based on the normal distribution quantiles. As described above, in some embodiments, a threshold is set for optimal trade-off between the detection of true positives and rate of false negative results. Namely, the threshold is chosen to maximize the sum oftrue positives and true negatives, or minimize the sum ofthe false positives and false negatives.
Certain embodiments provide a method for providing prenatal diagnosis of a fetal chromosomal aneuploidy in a biological sample comprising fetal and maternal nucleic acid les. The diagnosis is made based on obtaining sequence ation from at least a portion of the mixture of the fetal and maternal nucleic acid molecules derived from a ical test sample, e.g., a maternal plasma sample, computing from the sequencing data a normalizing chromosome dose for one or more chromosomes of st, and/or a normalizing segment dose for one or more ts of interest, and determining a statistically significant difference between the chromosome dose for the chromosome of interest and/or the segment dose for the segment of interest, respectively, in the test sample and a threshold value established in a plurality of qualified (normal) samples, and providing the prenatal diagnosis based on the statistical difference. As described in step 160 of the method, a sis of normal or ed is made. A "no call" is ed in the event that the diagnosis for normal or affected cannot be made with confidence.
In some embodiments, two thresholds can be chosen. A first old is chosen to minimize the false positive rate, above which samples will be classified as "Affected", and a second threshold is chosen to minimize the false negative rate, below which samples will be classified as "unaffected". Samples having NCVs above the second threshold but below the first threshold can be classified as "Aneuploidy ted" or "No call" samples, for which the presence or absence of aneuploidy can be confirmed by independent means. The region n the first and second thresholds can be ed to as a "no call" region.
[00227] In some embodiments, the suspected and no call thresholds are shown in Table 1. As can be seen, the thresholds of NCV vary across different chromosomes. In some embodiments, the thresholds vary according to the FF for the sample as explained above. Threshold techniques applied here contribute to improved sensitivity and ivity in some embodiments.
TABLE 1. Suspected and Affected NCV Thresholds Bracketing No-Call Ranges Suspected Affected Chr 13 3.5 4.0 Chr 18 3.5 4.5 Chr 21 3.5 4.0 Chr X (XO, XXX) 4.0 4.0 Chr Y (XX vs XY) 6.0 6.0 Fragment Size and Sequence Coverage Analyses
[00228] As mentioned above, fragment size parameters, as well as coverage, may be used to evaluate CNV. nt size of a cell free nucleic acid fragment, e.g., a cIDNA fragment may be obtained by pair end sequencing, electrophoresis (e.g., microchip-based capillary electrophoresis), and other methods known in the art.
Figure 2A thematically illustrates how paired end sequencing may be used to determine both fragment size and sequence coverage.
[00229] The top half of Figure 2A a shows a diagram of a fetal cell free DNA fragment and a maternal cell free DNA fragment providing a template for a paired end sequencing process. Conventionally, long nucleic acid sequences are fragmented into shorter ces to be read in a paired end sequencing process. Such fragments are also referred to as s. Fragmenting is unnecessary for cell free DNA because they already exist in fragments mostly shorter than 300 base pairs. It has been shown that fetal cell free DNA nts in maternal plasma are longer than maternal cell free DNA fragments. As shown at the top of figure 2A, cell free DNA of fetal origin have an average length of about 167 base pairs, while cell free DNA of maternal origin have an average length of about 175 base pairs. In paired end sequencing on certain rms, such as the Illumina's sequencing by synthesis platform as described further hereinafter, adaptor ces, index sequences, and/or prime sequences are ligated to the two ends of a fragment (not shown in Figure 2A). A fragment is first read in one direction, ing read 1 from one end ofthe fragment.
Then a second read starts from the opposite end of the fragment, providing the rea 2 sequence. The correspondence between read 1 and read 2 can be identified by their coordinates in the flow cell. Then read 1 and read 2 are mapped to a reference sequence as a pair of tags that are near each other, as shown in the bottom half of Figure 2A. In some embodiments, if the reads are long enough, the two reads can overlap in middle portion of the insert. After the pair is aligned to the reference ce, the relative distance between the two reads and the length of the fragment can be determined from the positions of the two reads. Because paired end reads provide twice as many base pairs as single end reads of the same read length, they help to improve alignment ies, especially for sequences with many repeats or non-unique sequences. In many embodiments, a reference ce is divided into bins, such as 100 K base pair bins. After paired end reads are aligned to the reference sequence, the number of reads d to a bin can be determined. The number as well as the lengths of s (e.g., cIDNA fragments) can also be determined for a bin. In some embodiments, if an insert straddles two bins, half of an insert may be attributed to each bin.
Figure 2B shows an embodiment providing process 220 for using sizebased coverage to determine a copy number variation of a nucleic acid sequence of st in a test sample including cell-free nucleic acid fragments originating from two or more genomes. As disclosed herein, a ter is "biased toward a fragment size or size range" when: I) the parameter is favorably weighted for the fragment size or size range, e.g., a count weighted more heavily when associated with fragments of the size or size range than for other sizes or ranges; or 2) the parameter is obtained I0 from a value that is favorably weighted for the fragment size or size range, e.g., a ratio obtained from a count weighted more heavily when associated with fragments of the size or size range. A nt size or size range may be a characteristic of a genome or a portion thereof when the genome produces nucleic acid fragments enriched in or having a higher concentration of the size or size range relative to nucleic acid fragments from another genome or another portion ofthe same genome.
Process 220 starts by receiving sequence reads ed by sequencing the cell-free nucleic acid fragments in the test sample. See block 222. The two or more s in the test sample may be a genome of a pregnant mother and a genome of a fetus carried by the pregnant mother. In other ations, the test sample includes cell free DNA from tumor cells and cted cells. In some embodiments, because of the high signal to noise ratio provided by the size-biased coverage, the sequencing of the cell free nucleic acid fragments are performed without the need to amplify the nucleic acid fragments using PCR. Process 200 further involves aligning the sequence reads of the cell-free c acid fragments to a reference genome that includes the sequence of interest and is divided into a ity of bins. sful alignment results in test sequence tags, which include sequence and its location on the reference sequence. See block 224. Then process 220 proceeds by determining sizes of the cell-free nucleic acid nts existing in the test sample. Some embodiments applying paired end sequencing provide the length of an insert associated with a sequence tag. See block 226. The terms "size" and "length" are used interchangeably when they are used with reference to nucleic acid sequences or nts. In the embodiment illustrated here, s 220 further involves weighting the test sequence tags based on the sizes of cell-free nucleic acid fragments from which the tags are obtained. See block 228. As used herein, "weighting" refers to modifying a quantity using one or more les or functions.
The one or more variables or functions are considered a "weight." In many embodiments, the variable is multiplied by the weight. In other embodiments, the variable may be modified exponentially or otherwise. In some embodiments, weighting the test ce tags is performed by biasing the ges toward test sequence tags obtained from cell-free nucleic acid fragments of a size or a size range characteristic of one genome in the test sample. As sed herein, a size is teristic of a genome when the genome has an enriched or higher concentration of nucleic acid of the size relative to another genome or another portion of the same genome.
In some embodiments, weighting function may be a linear or nonlinear function. Examples of applicable non-linear functions include, but are not limited to Heaviside step functions, box-car functions, case functions, or sigmoidal functions. In some embodiments, a Heaviside function or a box-car on is used, such that a tag in a specific size range is multiplied by a weight of 1, and tags outside of the range is multiplied by a weight of 0. In some ments, fragments between 80 and 150 base pairs are given a weight of 1, while fragments outside of this range is given a weight of 0. In these examples, the weighting is discreet, being zero or one depending on whether the parameter of all the value falls inside or outside a ular range. atively, weights are calculated as a continuous function of the fragment size or other aspect of the ated parameter value.
In some embodiments, the weights for nts in one size range are positive, and those in another range are negative. This may be used to help enhance signal when the directions of the difference between two genomes have the opposite signs. For instance, read counts have a weight of 1 for 80-150 base-pair insert, and a weight of -1 for 160-200 base-pair insert.
Weighs may be given to counts, as well as other parameters. For instance, weighting may also be applied to the fractional or ratio parameters that use fragment size. For example, the ratio may give fragments in certain sub-ranges greater weight than fragments and other size bins.
Then coverages are calculated for the bins based on the weighted test sequence tags. See block 230. Such coverages are considered size-biased. As explained above a value is biased toward a fragment size or size range when the parameter is favorably weighted for the nt size or size range. Process 200 further involves identifying a copy number variation in the sequence of interest from the calculated coverages. See block 232. In some embodiments, as further explained hereinafter in connection with Figures 2C, 3A-3K, and 4, the coverages may be adjusted or corrected to remove noise in the data, thereby increasing the -tonoise ratio. In some applications, the coverage based on the weighted tags obtained in s 220 provides both a higher ivity and/or a higher selectivity compared to un-weighted coverages in determining the copy number variation. In some ations, the example workflow provided below can further improve the sensitivity and selectivity for CNV analysis.
Workffow Example for Analyzing Fragment Size and/or Sequence Coverage Some embodiments disclosed provide methods to determine ce coverage quantities with low noise and/or high signal, providing data to determine various genetic conditions related to copy number and CNV with improved sensitivity, selectivity, and/or efficiency ve to sequence coverage quantities obtained by conventional methods. In certain embodiments, sequences from a test sample are processed to obtain sequence coverage quantities.
The process makes use of certain information available from other sources. In some implementations, all of this information is obtained from a training set of s known to be cted (e.g., not aneuploid). In other embodiments, some or all of the information is obtained from other test samples, which may be provided "on-the-fly" as multiple samples are analyzed in the same process.
] In n ments, sequence masks are employed to reduce data n01se. In some ments, both the sequence of interest and its normalizing sequences are masked. In some embodiments, different masks may be employed when different chromosomes or segments ofinterest are considered. For example one mask (or group of masks) may be employed when chromosome 13 is the chromosome of interest and a different mask (or group of masks) may be employed with chromosome 21 is the chromosome of interest. In certain embodiments, the masks are defined at the resolution of bins. Therefore, in one example, the mask resolution is 100 kb. In some embodiments, a distinct mask may be d to chromosome Y.
The masked exclusion regions for chromosome Y may be provided at a finer resolution (1 kb) than for other chromosomes of interest, as bed in US Provisional Patent Application No. 61/836,057, filed June 17, 2013 [attorney docket no. ARTEP008P]. The masks are ed in the form of files fying excluded genomic reg10ns.
] In certain embodiments, the process utilizes an expectation value of normalized coverage to remove bin-to-bin variation in the profile of a sequence of interest, which variation is uninformative for ination of CNV for the test sample. The process adjusts normalized coverage quantities according to the expectation value of normalized coverage for each bin across the entire genome, or at least the bins ofthe robust chromosomes in the reference genome (for use in operation 317 below). Parameters other than coverage may be improved by this process as well.
The expectation value may be determined from a ng set of unaffected samples.
As an example, the expectation value may be a median value across the training set samples. The expected coverage values of the samples may be determined as the number of unique dundant tags aligned to a bin divided by the total number of unique non-redundant tags aligned to all bins in the robust chromosomes of the reference genome.
Figure 2C s a flowchart of a process 200 for determining a fragment size parameter for a sequence of interest, which parameter is used to evaluate the copy number of the ce of interest in a test sample in block 214.
This process removes systematic variation common across unaffected training samples, which variation increases noise in the analysis for CNV evaluation. It also removes GC bias ic to a test sample, thereby increasing the signal-to-noise ratio in data analysis. It is worth noting that process 200 may also be applied to coverage, regardless of if the coverage is biased by size or not. Similarly, the processes in Figures 2D, 3, and 4 are equally applicable to ge, fragment size weighted coverage, nt size, fraction or ratio of fragments in a defined size range, methylation level offragments, etc.
The process 200 starts by ing sequence reads of the test sample as indicated in block 202. In some ments the sequence reads are obtained by cing DNA segments obtained from a pregnant woman's blood including cIDNA ofthe mother and the fetus. The process proceeds to align the sequence reads to a nce genome including the ce ofinterest, providing test sequence tags.
Block 204. In some ments, reads that are aligned to more than one site are excluded. In some embodiments multiple reads align to the same site are excluded or reduced to a single read count. In some embodiments, reads aligned to ed sites are also excluded. Therefore, in some embodiments, only the uniquely aligned, non- redundant tags aligned to non-excluded sites are d to provide a non-excluded site count (NES count) for determining the coverage or other parameters of each bin.
Process 200 provides sizes of the cell-free nucleic acid fragments existing in the test sample. In some embodiments using paired end sequencing, an insert size/length can be obtained from the locations of a pair of reads at the ends of the insert. Other techniques can be used to determine fragment size. See block 205.
Then, in bins of the reference genome, including bins in the sequence of interest, process 200 determines values of a fragment size parameter biased toward nt sizes characteristic of one ofthe genomes. The term "fragment size parameter" refers to a parameter that relates to the size or length of a fragment or a collection of fragments of nucleic acid fragments; e.g., cIDNA nts obtained from a bodily fluid. As used herein, a parameter is "biased toward a fragment size or size range" when: 1) the parameter is favorably weighted for the fragment size or size range, e.g., a count weighted more heavily when associated with fragments of the size or size range than for other sizes or ranges; or 2) the parameter is obtained from a value that is favorably weighted for the nt size or size range, e.g., a ratio obtained from a count weighted more heavily when associated with fragments of the size or size range. A fragment size or size range may be a characteristic of a genome or a portion thereof when the genome produces nucleic acid fragments enriched in or having a higher concentration of the size or size range relative to nucleic acid nts from another genome or another portion ofthe same .
In some embodiments, the fragment size parameter is a size-weighted count. In some embodiments a fragment is weighted 1 in a range, and 0 outside ofthe range. In other embodiments, the fragment size parameter is a fraction or a ratio of WO 36059 fragments in a size range. See block 206. In some embodiments, the value of the fragment size parameter (or coverage, as noted above) of each bin is divided by the value of the parameter of the normalizing sequence in the same sample, providing a normalized parameter.
[00244] s 200 then es a global profile of the sequence of interest.
The global profile comprises an expected parameter value in each bin obtained from a training set of unaffected training samples. Block 208. Process 200 removes variation common in the training sample by adjusting the normalized parameter values of the test sequence tags ing to the ed parameter values to obtain a global- profile-corrected values of the parameter for the sequence of interest. Block 210. In some embodiments, the expected value ofthe parameter obtained from the training set provided in block 208 is a median of across the training samples. In some embodiments, operation 2010 adjusts the normalized value of the parameter by subtracting the expected value of the ter from the normalized value of the parameter. In other embodiments, ion 210 s the normalized value of the ter by the expected value of the parameter of each bin to produce globalprofile corrected value ofthe parameter.
In addition to or instead of global profile correction, process 200 removes GC bias specific to the test sample by adjusting the parameter value. As shown in block 212, the s adjusts the global-profile-corrected parameter value based on the relation between GC content level and the global-profile-corrected coverage existing in the test , thereby obtaining a sample-GC-corrected value ofthe nt size parameter. After adjusting for systematic variation common in the unaffected training samples and within-subject GC bias, the process provides fragment size value corrected for global profile and/or GC variance, which value is used to evaluate CNV of the sample with improved sensitivity and specificity. In some implementations, the fragment size value may be adjusted using a principal component analysis method to remove components of variance unrelated to copy number variation of the sequence of interest as further described with reference to block 719 of Figure 2F. In some implementations, the fragment size value may be curated by removing outlier bins of within a sample as described with nce to block 321 ofFigure 3A.
Multi-pass Process for Copy Number Determination Using Multiple Parameters As emphasized above, the processes disclosed herein are suitable for determining CNV using multiple parameters, including but not limited to coverage, fragment size ed coverage, fragment size, fraction or ratio of fragments in a defined size range, methylation level of nts, etc. Each ofthese parameters may be separately processed to individually contribute to a final copy number variation determination.
In some embodiments, similar processes may be applied to a size- weighted coverage analysis and a fragment size analysis, both of which are fragment size parameters. Figure 2D shows a flow chart of two overlapping passes of work flow 600, pass 1 for size-weighted coverage, and pass 2 for fragment size analysis. In another embodiment not shown here, methylation level can be processed in one additional pass. The two passes can include comparable operations to obtain adjusted coverage information, on which determination of CNV is based.
An initial single pass n of the process starts by rece1vmg sequencing data, see block 602, and continues through computing counts as bed above, see block 612. After this point, the ed process splits into two passes, as described above. ing to the initial portion of the process, the workflow converts sequencing data into sequence reads. When the sequencing data is derived from multiplex sequencing, the sequence reads are also de-multiplexed to identify the source ofthe data. See block 604. The sequence reads are then aligned to a reference sequence, where the aligned ce reads are provided as sequence tags. See block 606. Then sequence tags are filtered to obtain non-excluded sites (NESs), which are unambiguously mapped, non-duplicated ce tags. Sequence tags are organized into bins of specific sequence length, such as 1 kb, 100 kb, or 1 Mb. See block 610. In some embodiments involving is of syndrome specific regions, the bins are 100 kb. In some ments, bins exhibiting high variability may be masked using a sequence mask obtained from a plurality of unaffected samples in a manner as described in Figure 3A, block 313. Then the tags in the NESs are counted to e coverages to be normalized and adjusted for analysis of CNV. See block 612.
In the ed ment, ions 604, 606, 610, and 612 are performed once and most ofthe remaining operations are performed twice, once for a size-weighted coverage analysis (pass 1) and once for a fragment size is (pass 2). In other embodiments, one or more ofthe operations shown as being performed in two passes are performed only once and the results are shared in both processes.
Examples of such shared operations include operations 614, 616, and 618.
In the depicted embodiments, the obtained coverages (size weighted counts) or fragment size parameter (size fractions or ratios) of NESs are normalized by, e.g., dividing the value NES of a bin by the total NESs of the genome or a set of normalizing chromosomes. In some embodiments, only the coverage is normalized, while the fragment size parameter does not need to be normalized, because it is not affected by sequencing depth the same way as coverage. See block 614. Then, in some embodiments, the variance common to a training set including unaffected samples is removed, which variance is unrelated to the CNV of st. In the depicted embodiment, the common variance is represented as a global wave profile obtained from unaffected samples in the manner similar to the global wave profile described above. In some embodiments as illustrated in Figure 6, the unaffected samples used to obtain a global wave profile include samples coming from the same flow cell or processing batch. See block 616. The calculation ofthe flow cell specific global wave is further explained hereinafter. In the depicted embodiment, after the global wave profile has been d, ges are corrected for GC level on a -specific basis. See block 616. Some algorithms for GC correction are described in further details hereinafter in the text associated with Figure 3A, block 319.
[00251] In the depicted embodiment, in both pass 1 for weighted coverage analysis and pass 2 for fragment size analysis, data may be further filtered for noise specific to an dual sample, e.g., data of outlier bins that have coverages extremely ent from other bins may be d from analysis, which difference cannot be attributed to the copy number variation of interest. See block 622. This within-sample filtering operation may pond to block 321 in Figure 3A.
In some ments, after single sample filtering, the weighted coverage values of pass 1 and the fragment size parameter of pass 2 are both enriched in target signal over reference. See blocks 624 and 628. Then, the coverage and the fragment size parameter for the chromosome each is used to ate a chromosome dose and a normalized chromosome value (NCV) as described above. The NCV then may be compared to a criterion to determine a score ting a probability of a CNV. See blocks 626 and 630. The scores from the two passes can then be combined to provide a composite, final score, which determines whether an aneuploidy should be . In some embodiments, the scores of 626 and 630 are ttest statistics or Z values. In some embodiments, the final score is a chi square value.
In other embodiments, the final score is a root mean square of the two t values or z scores. Other means to combine the two scores from the two paths may be used to improve the overall sensitivity and selectivity in CNV detection. Alternatively, one may e the two scores from the two passes by logical operations, e.g., AND operation or OR ion. For instance, when a high sensitivity is preferred to ensure low false negative, a CNV call can be made when the score from pass 1 OR pass 2 meets a call criterion. On the other hand, if high selectivity is desired to ensure low false positive, a CNV call can be made only ifthe score from both pass 1 AND pass 2 meet a call criterion.
It is notable that there is a trade-off n ivity and selectivity using such logical operations above. In some embodiments, a two-step sequencing approach is applied to overcome the trade-off as further described hereinafter.
Briefly, the l scoring of a sample is ed against a relatively low, first threshold designed to increase sensitivity, and if the sample scores higher than the first threshold, it undergoes a second round of sequencing, which is deeper than the first one. Such a sample is then re-processed and ed in a workflow similar to that described above. Then the resulting score is compared to a relatively high, second threshold designed to e the sensitivity. In some embodiments, the samples undergoing a second round of sequencing score relatively low among those that score above the first threshold, thereby reducing the number of samples that need to be resequenced.
In some embodiments, a 3rd pass usmg a 3rd parameter can be employed. One example of this 3rd pass is methylation. The methylation may be determined directly through measuring the methylation of the nucleic acids from the sample or indirectly as a parameter that ates with fragment size of the cell free nucleic acids.
In some embodiments, this 3rd parameter is a 2nd coverage or count based parameter, where the counts are based on nt sizes outside the primary fragment size used in the first count based parameter. When fragments between 80 and 150 base pairs are used for generating the count or coverage parameter, they exclude about 70% of the reads from a sequencing. To the extent that these excluded reads still have some potentially useful signal, they may be used in a 3rd parameter which includes the excluded reads or reads in a size-based fraction that is outside of or overlaps with the size-based on used in the first parameter. In this regard, the reads and associated coverage values taken from the excluded fragments may be given a lower weight. In other words, the copy number variation parameter ated using these reads may be ed less importance in making a final copy number variation call. Alternatively, as described above, the tags outside of the size range in the first parameter may take on a negative value when the two genomes have opposite characteristics in the two size .
[00256] In various implementations, the coverages in processes 200, 220, and 600 are biased toward tags from fragments at a shorter end of a fragment size spectrum. In some embodiments, the coverages are biased toward tags from fragments of sizes shorter than a specified value. In some embodiments, the ges are biased toward tags from fragments in a range of fragment sizes, and the upper end ofthe range is about 150 base pairs or fewer.
In various implementations of processes 200, 220, and 600, the sequence reads are obtained by sequencing the cell-free nucleic acid fragments without first using PCR to amplify nucleic acids of the ree nucleic acid fragments. In various embodiments, the sequencing reads are obtained by sequencing the cell-free nucleic acid fragments to a depth of no greater than about 6 M fragments per sample. In some embodiments, the cing depth is no greater than about 1 M fragments per sample. In some embodiments, sequencing reads are obtained by lex sequencing, and the number of s multiplexed is at least about 24.
] In various implementations of processes 200, 220, and 600, the test sample comprises plasma from an individual. In some embodiments, the ses further comprising obtaining the cell-free nucleic acid from the test sample. In some embodiments, the processes further comprising sequencing the cell-free nucleic acid fragments originating from two or more genomes.
In various implementations of processes 200, 220, and 600, the two or more s compnse genomes from a mother and a fetus. In some implementations, the copy number variation in the sequence of interest comprises aneuploidy in the genome ofthe fetus.
[00260] In some implementations of processes 200, 220, and 600, the two or more s comprise genomes from cancer and somatic cells. In some implementations, the processes comprising using a copy number variation in the cancer genome to diagnose cancer, monitor the progress of cancer, and/or determine a treatment for cancer. In some implementations, the copy number variation causes a genetic abnormality.
] In some implementations of ses 200, 220, and 600, the coverages are biased toward tags from nts at a longer end of a fragment size spectrum. In some implementations, the coverages are biased toward tags from fragments of sizes longer than a specified value. In some implementations, coverages are biased toward tags from fragments in a range of fragment sizes, and wherein the lower end ofthe range is about 150 base pairs or more.
In some implementations of processes 200, 220, and 600, the processes further involves: determining, in bins ofthe reference genome, including the ce of interest, levels of methylation of the ree nucleic acid fragments in said bins, and using the levels of methylation, in on to or instead of the calculated coverages or the values of the fragment size parameter to identify a copy number variation. In some implementation, using the methylation levels to identify a copy number variation involves providing a global methylation profile for the bins of the sequence of interest. The global methylation profile includes expected levels of methylation in at least bins of the sequence of st. In some entations, the ed levels of methylation are obtained from lengths of cell-free nucleic acid fragments in a training set of unaffected training samples comprising nucleic acids sequenced and aligned in substantially the same manner as the nucleic acid fragments of the test sample, the expected levels of methylation exhibiting variation from bin to bin. In some entations, the processes involve adjusting the value ofthe levels of ation using the expected levels of methylation in the bins of at least the sequence of interest, thereby obtaining global-profile-corrected values of the levels of methylation for the sequence of interest. the processes further involve identifying a copy number variation usmg -profile-corrected coverages and the globalprofile-corrected levels of ation. In some implementations, identifying a copy number variation using the global-profile-corrected coverages and the global-profilecorrected levels of methylation further comprises: adjusting the global-profile- s ted coverages and the global-profile-corrected levels of methylation based on GC content levels, thereby obtaining GC-corrected coverages and GC-corrected values ofthe levels of methylation for the sequence of interest; and identifying a copy number variation using the rected coverages and the GC-corrected levels of methylation.
[00263] In some implementations of processes 200, 220, and 600, the fragment size parameter comprises a fraction or ratio including a portion ofthe cell-free nucleic acid fragments in the test sample having fragment sizes shorter or longer than a threshold value. In some implementations, the fragment size parameter es a fraction including (i) a number ments in the test sample within a first size range including 110 base pairs, and (ii) a number offragments in the test sample within a second size range comprising the first size range and sizes outside the first size range.
Copy Number Determination Using a Three-pass Process, Likelihood Ratios, T tics, and/or Fetal ons ] Figure 2E shows a flow chart of a three-pass process for evaluating copy number. It includes three overlapping passes of work flow 700, which includes pass 1 (or 713A) analysis of coverage of reads associated with fragments of all sizes, pass 2 (or 713B) analysis of coverage of reads associated with shorter fragments, and pass 3 (or 713C) is ofrelative frequency of shorter reads relative to all reads.
Process 700 is similar to process 600 in its overall organization.
Operations indicated by blocks 702, 704, 706, 710, 712 may be performed in the same or a similar manner to operations indicated by blocks 602, 604, 606, and 610, and 612. After read counts are obtained, coverage is determined using reads from fragments of all sizes in pass 713A. Coverage is determined using reads from short fragments in pass 713B. Frequency of reads from short fragments relative to all reads is determined in pass 713C. The relative ncy is also referred to as a size ratio or a size fraction elsewhere . It is an example of a fragment size characteristic. In some implementations, short fragments are fragments shorter than about 150 base 2016/067886 pairs. In various implementations, short fragments can be in the size ranges of about 50-150, 80-150, or 110-150 base pairs. In some implementations, the third pass, or pass 713C, is optional.
The data of the three passes 713A, 713B, and 713C all undergo ization operations 714, 716, 718, 719, and 722 to remove variance unrelated to copy number ofthe sequence ofinterest. These ization operations are boxed in blocks 723. Operation 714 involves normalizing the analyzed quantity of the sequence of interest by dividing the analyzed quantity by the total value of the ty ofthe reference sequence. This normalization step uses values obtained from a test sample. Similarly, operations 718 and 722 normalize the ed quantity using values obtained from the test sample. ions 716 and 719 use values obtained from a training set ofunaffected samples.
Operation 716 removes variance of a global wave obtained from the training set of cted samples, which uses the same or similar methods as described with reference to block 616. Operation 718 removes variance ofindividualspecific GC variance using the same or similar manner methods as described with reference to block 618. ion 719 removes further variance using a principal component analysis (PCA) method. The variance removed by the PCA methods is due to factors unrelated to copy number of the sequence of interest. The analyzed quantity in each bin (coverage, fragment size ratio, etc.) provides an ndent variable for the PCA, and the samples of the unaffected training set supply values for these independent variables. The samples of the training set all include samples having the same copy number of the sequence of st, e.g., two copies of a somatic chromosome, one copy of the X chromosome (when male samples are used as cted samples), or two copies of the X chromosome (when female s are used as unaffected samples). Thus, variance in the samples does not result from an aneuploidy or other difference in copy number. The PCA of the training set yields principal components that are unrelated to copy number of the sequence of interest. The principal components can then be used to remove variance in a test sample unrelated to the copy number ofthe sequence ofinterest.
In n embodiments, the variance of one or more of the principal components is removed from the test sample's data using the coefficients ted from unaffected samples' data in a region outside of the sequence of st. In some implementations, the region represents all robust chromosomes. For instance, a PCA is performed on normalized bin coverage data of training normal samples, thereby providing principal components corresponding to dimensions in which most variance in the data can be captured. Variance so captured is ted to copy number variation in the sequence of interest. After the principal ents have been obtained from the training normal samples, they are applied to test data. A linear regression model with test sample as response variable and principal components as dependent les is generated across bins from a region outside of the sequence of interest. Resulting regression coefficients are used to normalize the bin coverage of the region of interest by subtracting the linear combination of principal components defined by the estimated regression coefficients. This removes variance unrelated to CNV from the sequence of interest. See block 719. The residual data is used for ream is. Additionally, operation 722 removes outlier data points using methods described with reference to block 622.
After undergoing the normalization operations m block 723, the coverage values of all bins have been "normalized" to remove sources of variation other than aneuploidy or other copy number variations. In a sense, the bins of the ce of interest are enriched or altered relative to other bins for purposes of copy number variation detection. See block 724, which is not an operation but represents the ing coverage values. The normalization operations in large block 723 may increase the signal and/or reduce the noise of the quantity under is. Similarly, the coverage values of short fragments for the bins have been normalized to remove sources of variation other than aneuploidy or other copy number variations as shown in block 728, and the relative frequency of short fragments (or size ratio) for the bins have been similarly normalized to remove sources of variation other than aneuploidy or other copy number variations as shown in block 732. As with block 724, blocks 728 and 732 are not operations but represents the coverage and relative frequency values after the processing large block 723. It should be tood, that the ions in large block 723 may be modified, rearranged, or removed. For example, in some embodiments, PCA operation 719 is not performed. In other embodiments, the correcting for GC operation 718 is not performed. In other embodiments, the order of the operations is changed; e.g., PCA operation 719 is performed prior to correct for GC operation 718, The coverage of all fragments after normalization and vanance removal shown in block 724 is used to obtain tistic in block 726. rly, the coverage of short fragments after normalization and variance l shown in block 728 is used to obtain a t-statistic in block 730, and the ve frequency of short fragments after normalization and variance removal shown in block 732 is used to obtain at-statistic in block 734.
[00272] Figure 2F demonstrates why ng a t-statistic to copy number analysis can help to improve the accuracy of the analysis. Figure 2F shows, in each panel, the frequency distributions ofnormalized bin coverage of a sequence of interest and a reference sequence, with the sequence of interest distribution overlapping and obscuring the reference sequence distribution. In the top panel, bin ge for a sample having higher coverage is shown, having over 6 million reads; in the bottom panel, bin coverage for a sample having lower coverage is shown, having fewer than 2 million reads. The horizontal axis indicates coverage normalized relative to the mean coverage of the reference ce. The vertical axis indicates ve probability density related to numbers ofbins having the mean coverage values. Figure 2F is thus a type of histogram. The distribution for the sequence of interest is shown to the front, and the distribution of the reference sequence is shown to the back. The mean for the distribution of the sequence of interest is lower than that of the reference sequence, indicating a lowered copy number in the . The mean difference between the sequence of interest and the reference sequence is similar for the high coverage sample in the top panel and the low coverage sample in the bottom panel.
Thus, the difference in mean may, in some implementations, be used to identify a copy number variation in the sequence of interest. Note that the distributions of the high coverage sample have variances r than those of the low coverage sample.
Using only the mean to distinguish the two distributions does not capture the difference between the two distributions as well as using both mean and variance. At- statistic can reflect both the mean and variance ofthe bution.
In some implementations, operation 726 calculates a t-statistic as follows: t = X1-Xz where x1 is the bin coverage of the sequence of interest, x2 being the bin coverage of the reference region/sequence, s1 being the rd deviation of the coverages of the sequence of interest, s2 being the standard ion ofthe coverages ofthe reference region, n1 being the number ofbins ofthe sequence of interest; and n2 being the number ofthe bins ofthe nce region.
In some implementations, the reference reg10n includes all robust chromosomes (e.g., chromosomes other than those most likely to harbor an aneuploidy). In some implementations, the reference region includes at least one chromosome outside of the sequence of interest. In some imitations, the reference region includes robust chromosomes not including the sequence of st. In other implementations, the reference region es a set of chromosomes (e.g., a subset of chromosomes selected from the robust chromosomes) that have been determined to e the best signal detection ability for a set of training samples. In some embodiments, the signal detection ability is based on the ability of the reference region to discriminate bins harboring copy number variations from bins that do not harbor copy number variations. In some embodiments, the reference region is fied in a manner similar to that employed to determine a "normalizing sequence" or a "normalizing chromosome" as described in the section titled "Identification ofNormalizing Sequences." Returning to Figure 2E, one or more fetal fraction estimates (block 735) may be combined with any of the t tics in block 726, 730 and 734 to obtain a likelihood estimate for a ploidy case. See block 736. In some implementations, the one or more fetal fractions ofblock 740 are obtained by any of process 800 in Figures 2G, process 900 in Figure 2H, or process 1000 of Figure 21. The processes may be implemented in parallel using a ow as workflow 1100 in Figure 2J.
Figure 2G shows an e process 800 for determining fetal fraction from coverage information according to some entations of the disclosure.
Process 800 starts by obtaining coverage information (e.g., sequence dose values) of training samples from a training set. See block 802. Each sample ofthe training set is obtained from a pregnant woman known to be carrying a male fetus. Namely, the sample contains cIDNA of the male fetus. In some implementations, operation 802 may obtain sequence coverage ized in ways different from sequence dose as described , or it may obtain other coverage values.
Process 800 then involves calculating fetal fractions of the training samples. In some implementations, fetal fraction may be calculated from the sequence dose : FF. = _ 2 X Rxrmedian(Rxi) l median(Rxi) where Rxj is the ce dose for a male sample, median(Rxi) being the median of the sequence doses for female s. In other implementations, mean or other central tendency measures may be used. In some implementations, the FF may be ed by other methods, such as the relative ncy of X and Y chromosomes. See block 804.
Process 800 further involves dividing the reference sequence into multiple bins of subsequences. In some implementations, the reference sequence is a complete genome. In some implementations, the bins are 100 kb bins. In some implementations, the genome is divided into about 25,000 bins. The process then obtains coverages of the bins. See block 806. In some implementations, the coverages used in block 806 are obtained after undergoing normalizing ions shown in block 1123 of Figure 2J. In other implementations, coverages from different size range may be used.
Each bin is associated with coverages ofthe samples in the training set.
Therefore, for each bin a correlation may be obtained between the coverage of the s and the fetal fractions of the samples. Process 800 involves obtaining correlations between fetal fraction and coverage for all the bins. See block 808. Then the process selects the bins having correlation values above a threshold. See block 810. In some implementations, bins having the 6000 highest correlation values are selected. The purpose is to identify bins that demonstrate high correlation between coverage and fetal fraction in the training samples. Then the bins may be used to predict fetal fraction in the test sample. Although the training samples are male samples, the ation between fetal fraction and coverage may be generalized to male and female test samples.
Using the selected bins having high correlation values, the process obtains a linear model relating fetal fraction to coverage. See block 812. Each selected bin provides an independent variable for the linear model. Therefore, the obtained linear model also includes a ter or weight for each bin. The weights of the bins are adjusted to fit the model to the data. After obtaining the linear model, s 800 involves applying coverage data of the test sample to the model to determine the fetal fraction for the test sample. See block 814. The d coverage data of the test sample are for the bins that have high correlations between fetal fraction and coverage.
[00285] Figure 2J shows workflow 1100 for processmg sequence reads information of which can be used to obtain fetal fraction estimates. The ow 1100 shares similar processing steps as workflow 600 in Figure 2D. Blocks 1102, 1104, 1106, 1110, 1112, 1123, 1114, 1116, 1118, and 1122 respectively correspond to blocks 602, 604, 606, 610, 612, 623, 614, 616, 618, and 622. In some implementations, one or more normalizing operations in the 123 block are optional.
Pass 1 provides coverage information, which may be used in block 806 ofprocess 800 shown in Figure 2G. Process 800 then can yield a fetal fraction te 1150 in Figure 2J.
In some implementations, a plurality of fetal fraction estimates (e.g., 1150 and 1152 in Figure 2J) may be combined to e a composite fetal fraction estimate (e.g., 1154). Various methods may be used to obtain fetal fraction estimates.
For instance, fetal fraction may be obtained from ge information. See block 1150 of Figure 2J and process 800 of Figure 2G. In some implementations, fetal fraction can also be estimated from size distribution of fragments. See block 1152 of Figure 2J and process 900 of Figure 2H. In some implementations, fetal fraction can also be estimated from 8-mer frequency distribution. See block 1152 ofFigure 2J and process 1000 ofFigure 21.
In a test sample including cIDNA of male fetus, fetal fraction may also be estimated from the ge of the Y chromosome and/or the X chromosome. In some implementations, a composite estimate of fetal fraction (see, e.g., block 1155) for a putatively male fetus is obtained by using information selected from the group consisting of: a fetal on ed from coverage information of bins, a fetal fraction obtained from fragment size ation, a fetal fraction obtained from coverage ofthe Y chromosome, a fetal on ed from the X chromosome, and any combinations f. In some implementations, the putative sex of the fetus is obtained by using the ge of the Y chromosome. Two or more fetal fractions (e.g., 1150 and 1152) may be combined in s ways to provide a composite estimate of fetal fraction (e.g., 1155). For instance, an average or a weighted average approach may be used in some implementations, wherein weighting can be based on the statistical confidence ofthe fetal fraction estimate.
In some implementations, a composite estimate of fetal fraction for a putatively female fetus is obtained by using information selected from the group consisting of: a fetal fraction obtained from coverage information of bins, a fetal fraction obtained from fragment size information, and any combinations thereof.
Figure 2H shows a process for determining fetal fraction from size distribution information according to some implementations. Process 900 starts by obtaining coverage information (e.g., sequence dose values) of male training samples from a training set. See block 902. Process 900 then es calculating fetal fractions of the training samples using methods described above with reference to block 804. See block 904. s 900 proceeds to divide a size range into a plurality of bins to provide fragment-size-based bins and determine frequencies of reads for the fragment-size-based bins. See block 906. In some implementations, the frequencies of nt-size-based bins are obtained without izing for factors shown in block 1123. See path 1124 ofFigure 2J. In some implementations, the ncies of fragment-size-based bins are obtained after ally undergoing normalizing operations shown in block 1123 of Figure 2J. In some implementations, the size range 1s divided into 40 bins. In some implementations, the bin at the low end includes fragments of size smaller than about 55 base pairs. In some implementations, the bin at the low end includes fragments of size in the range of about 50-55 base pairs, which excludes information for reads shorter than 50 bp. In some implementations, the bin at the high end includes fragments of size larger than about 245 base pairs. In some implementations, the bin at the high end includes fragments of size in the range of about 245-250 base pairs, which excludes information for reads longer than 250 bp.
Process 900 proceeds by obtaining a linear model relating fetal fraction to frequencies of reads for the nt-size-based bins, using data of the ng samples. See block 908. The obtained linear model includes independent variables for the frequencies of reads of the size-based bins. The model also includes a parameter or weight for each size-based bin. The weights of the bins are adjusted to fit the model to the data. After ing the linear model, process 900 involves ng read frequency data of the test sample to the model to determine the fetal fraction for the test sample. See block 910.
In some implementations, an 8-mer frequency may be used to calculate fetal fraction. Figure 2I shows an example process 1000 for ining fetal fraction from 8-mer frequency information according to some implementations of the disclosure. Process 1000 starts by obtaining coverage information (e.g., sequence dose values) of male training samples from a training set. See block 1002. s 1000 then involves calculating fetal fractions of the ng samples using any of the methods described for block 804. See block 1004.
Process 1000 further involves obtaining the frequencies of 8-mers (e.g., all possible permutations of 4 nucleotides at 8 positions) from the reads of each training sample. See block 1006. In some implementations, up to 65,536 or close to that many 8-mers and their frequencies are obtained. In some implementations, the frequencies of 8-mers are ed without normalizing for factors shown in block 1123. See path 1124 of Figure 2J. In some implementations, 8-mer frequencies are obtained after optionally undergoing izing operations shown in block 1123 of Figure 2J.
Each 8-mer is associated with frequencies of the samples in the training set. Therefore, for each 8-mer a correlation may be obtained between the 8- mer frequency of the samples and the fetal fractions of the samples. Process 1000 involves obtaining correlations between fetal fraction and 8-mer frequencies for all the . See block 1008. Then the process selects the 8-mers having correlation values above a threshold. See block 1010. The purpose is to fy 8-mers that demonstrate high correlation between 8-mer frequency and fetal fraction in the training s. Then the bins may be used to predict fetal fraction in the test sample. Although the training samples are male samples, the correlation between 2016/067886 fetal fraction and 8-mer frequency may be generalized to male and female test samples.
Using the selected 8-mers having high correlation values, the process s a linear model relating fetal fraction to 8-mer frequency. See block 1012.
Each ed bin provides an independent variable for the linear model. Therefore, the obtained linear model also es a parameter or weight for each bin. After obtaining the linear model, process I000 involves applying 8-mer frequency data of the test sample to the model to determine the fetal fraction for the test sample. See block 1014.
[00296] Returning to Figure 2E, in some implementations, process 700 involves obtaining a final ploidy hood in operation 736 using the istic based on the coverage of all nts provided by operation 726, the fetal fraction estimate ed by operation 726, and the t-statistic based on the coverage of the short fragments provided by operation 730. These implementations combine the results from pass I and pass 2 using a multivariate normal models. In some implementations for evaluating CNV, the ploidy likelihood is an aneuploidy likelihood, which is a likelihood of a model having an aneuploid assumption (e.g., trisomy or monosomy) minus the likelihood of a model having an euploid assumption wherein the model uses the t-statistic based on the coverage of all fragments, the fetal fraction estimate, and the t-statistic based on the coverage of the short fragments as an input and provides a likelihood as an output.
In some implementations, the ploidy likelihood is expressed as a likelihood ratio. In some implementations, likelihood ratio is modeled as: LR = Lfftotal q(ff total)*Pl(Tshort,Taulffest) Po(Tshort,Tau)
[00299] Wwhere p1 represents the likelihood that data come from a multivariate normal distribution representing a 3-copy or I-copy model, p0 represents the likelihood that data come from a multivariate normal distribution representing a 2- copy model, , Tan are T scores calculated from chromosomal coverage generated from short and all fragments, while tal) being the density distribution of fetal on (estimated from training data) considering the error associated with fetal fraction estimation. The model combine coverage generated from short fragments with ge generated by all fragments, which helps improving tion between coverage scores of affected and unaffected samples. In the ed ment, the model also makes use of fetal fraction, thereby further improves the y to discriminate between affected and unaffected samples. Here, the likelihood ratio is calculated using t-statistic based on coverage of all fragments (726), t-statistic based on coverage of short fragments (730), and a fetal on te provided by processes 800 (or block 726), 900, or I000 as described above. In some implementations, this likelihood ratio is used to analyze chromosomes 13, 18, and 21. some implementation, a ploidy likelihood obtained by operation 736 uses only the t-statistics obtained based on relative frequency of short fragments I0 provided by operation 734 of pass 3 and the fetal fraction estimate provided by operation 726, processes 800, 900, or 1000. The likelihood ratio may be calculated according to the following equation: LR = Lfftotal q(ff tota1)*P1 (Tshort_freq Iffest) Po (Tshort_freq) ] where p1 represents the likelihood that data come from a multivariate normal distribution representing a 3-copy or I-copy model, p0 represents the likelihood that data come from a multivariate normal distribution representing a 2- copy model, -freq is a T score calculated from relative frequency of short fragments, while q(fftotaL) being the density bution of fetal fraction (estimated from training data) considering the error associated with fetal fraction tion.
Here, the likelihood ratio is calculated using t-statistic based on relative ncy of short fragments (734) and a fetal fraction te provided by processes 800 (or block 726), 900, or I000 as described above. In some entations, this likelihood ratio is used to analyze chromosome X.
In some implementations, the likelihood ratio is calculated using t- statistic based on coverage of all fragments (726), t-statistic based on coverage of short nts (730), and relative frequency of short fragments (734). Moreover, fetal fraction obtained as describe above may be combined with t-statistics to calculate likelihood ration. By combining information from any of the three passes 713A, 713B, and 713C, the minative ability of the ploidy evaluation can be improved. See, e.g., Example 2 and Figure 12. In some implementations, different combinations may be used to obtain likelihood ratios for a chromosome, e.g., t statistics from all three passes, t statistics from the first and second passes, fetal on and three t-statistics, fetal fraction and one t statistic, etc. Then an l combination can be selected based on the models performance.
In some implementations for evaluating autosomes, the d likelihood ratio represents the likelihood of the d data having been obtained from a trisomy or monosomy sample relative to the likelihood of the modeled data having been obtained from a diploid sample. Such likelihood ratio may be used to determine trisomy or monosomy ofthe autosomes in some implementations.
In some implementations for evaluating the sex some, the likelihood ratio for monosomy X and the hood ratio for trisomy X are evaluated.
Moreover, a chromosome coverage measurement (e.g., CNV or coverage z score) for chromosome X and one for chromosome Y are also evaluated. In some implementations, the four values are evaluated using a decision tree to determine copy number of the sex chromosome. In some implementations, the decision tree allows determination of a ploidy case ofXX, XY, X, XXY, XXX, or XYY.
[00306] In some implementations, the likelihood ratio is transformed into a log likelihood ratio, and a criterion or threshold for calling an aneuploidy or a copy number variation can be empirically set to obtain a particular sensitivity and selectivity. For instance, a log likelihood ratio of 1.5 may be set for calling a trisomy 13 or a trisomy 18 based on a model's sensitivity and selectivity when d to a training set. Moreover, for instance, a call ion value of 3 may be set for a trisomy of some 21 in some applications.
Details ofan Exemplary Process for Determining Sequence Coverage Figure 3A presents an example of a process 301 for reducing the noise in sequence data from a test sample. Figures 3B-3J present data analyses at various stages of the process. This provides one example of a process flow that may be used in a multipass process such as depicted in Figure 2D. s 301 illustrated in Figure 3A uses ce tag coverage based on the number of sequence tags to evaluate copy . However, similar to the description above regarding process 100 for determining CNV with reference to Figure 1, other variables or parameters, such as size, size ratio, and methylation level, may be used instead of coverage for process 400. In some implementations, two or more variables can separately undergo the same process to derive two scores tive of probability of CNV, as shown above with reference to Figure 2D. Then the two scores may be combined to determine a CNV. Furthermore, coverage and other parameters may be weighted based on the size ofthe fragments from which tags are derived. For ease of reading, only coverage is referred to in process 300, but one should note that other parameters, such as size, size ratio, and methylation level, count weighted by size, etc. may be used in place of coverage.
As shown in Figure 3A, the depicted process begins with extraction of cIDNA from one or more s. See block 303. le extraction processes and apparatus are described elsewhere . In some embodiments, a process described in US Patent Application No. 61/801,126, filed March 15, 2013 (incorporated herein by reference in its entirety) extracts cIDNA. In some implementations, the apparatus processes cIDNA from multiple samples together to provide multiplexed libraries and sequence data. See blocks 305 and 307 in Figure 3A. In some embodiments, the tus processes cIDNA from eight or more test samples in parallel. As described elsewhere herein, a sequencing system may process extracted cIDNA to produce a library of coded (e.g., bar coded) cIDNA fragments. A sequencer sequences library of cIDNA to produce a very large number of sequence reads. Per sample coding allows demultiplexing of the reads in multiplexed samples. Each of the eight or more samples may have ds ofthousands or millions of reads. The process may filter the reads prior to additional operations in Figure 3A. In some embodiments, read filtering is a quality-filtering process enabled by re ms implemented in the sequencer to filter out erroneous and low quality reads. For example, Illumina' s Sequencing Control Software (SCS) and Consensus Assessment of Sequence and Variation software ms filter out ous and low quality reads by converting raw image data generated by the sequencing reactions into intensity scores, base calls, quality scored ents, and additional formats to provide biologically relevant ation for ream analysis.
After the sequencer or other apparatus generates the reads for a sample, an element ofthe system computationally aligns the reads to a reference genome. See block 309. Alignment is described elsewhere . The alignment produces tags, which contain read sequences with annotated location information specifying unique positions on the reference genome. In certain implementations, the system conducts a first pass alignment without regard for duplicate reads - two or more reads having identical sequences - and subsequently removes duplicated reads or counts duplicate reads as a single read to produce non-duplicated sequence tags. In other implementations, the system does not remove duplicated reads. In some ments, the process removes from consideration reads that are d to multiple locations on the genome to produce uniquely d tags. In some embodiments, uniquely aligned, non-redundant ce tags mapped to nonexcluded sites (NESs) are accounted for to yield non-excluded site counts (NES counts), which provide data to estimate coverage.
[00311] As explained elsewhere, excluded sites are sites found in regions of a reference genome that have been excluded for the purpose of counting sequence tags.
In some embodiments, excluded sites are found in regions of chromosomes that n repetitive sequences, e.g., centromeres and telomeres, and regions of chromosomes that are common to more than one chromosome, e.g., regions present on the Y-chromosome that are also present on the X chromosome. Non-excluded sites (NESs) are sites that are not excluded in a reference genome for the purpose of counting sequence tags.
Next, the system divides the aligned tags into bins on the reference genome. See block 311. The bins are spaced along the length of the nce genome. In some embodiments, the entire nce genome is divided into uous bins, which may have defined equal size (e.g., 100 kb). Alternatively, the bins may have a length determined dynamically, possibly on a per-sample basis. cing depth impacts optimal bin size selection. Dynamically sized bins may have their size determined by the library size. For example, the bin size may be determined to be the ce length required to accommodate 1000 tags, on average.
Each bin has a number of tags from a sample under consideration.
This number of tags, which reflects the "coverage" of the aligned ce, serves as a starting point for filtering and otherwise cleaning the sample data to reliably determine copy number variation in the sample. Figure 3A shows the cleaning operations in blocks 313 to 321.
In the embodiment depicted in Figure 3A, the process applies a mask to the bins of the reference genome. See block 313. The system may exclude WO 36059 ge in masked bins from consideration in some or all of the following process operations. In many cases, coverage values from masked bins are not ered any ofthe remaining operations in Figure 3A.
] In various implementations, one or more masks are applied to remove bins for regions of the genome found to exhibit high variability from sample to sample. Such masks are provided for both chromosomes of interest (e.g., chr13, 18, and 21) and other chromosomes. As explained elsewhere, a chromosome of interest is the chromosome under consideration as potentially harboring a copy number variation or other aberration.
] In some implementations, masks are identified from a training set of qualified samples using the following approach. Initially, each training set sample is processed and filtered according to operations 315 through 319 in Figure 3A. The ized and corrected coverage quantities are then noted for each bin and tics such as standard deviation, median absolute deviation, and/or coefficient of variation are calculated for each bin. Various filter combinations may be evaluated for each chromosome of interest. The filter combinations provide one filter for the bins ofthe chromosome ofinterest and a different filter for the bins of all other somes.
In some implementations, the choice of a izing chromosome (or group of chromosomes) is reconsidered after obtaining masks (e.g., by choosing cut- offs for a chromosome of interest as described above). After applying the sequence mask, the process of choosing a izing chromosome or chromosomes may be conducted as described elsewhere herein. For example, all possible combinations of chromosomes are evaluated as normalizing somes and ranked according to their ability to discriminate affected and unaffected samples. This process may (or may not) find a different optimal normalizing chromosome or group of chromosomes.
In other embodiments, normalizing chromosomes are those that result in the smallest variability in sequence dose for the sequence of interest across all ied s.
If a different normalizing chromosome or group of somes is identified, the process optionally executes the above described identification of bins to .
Possibly the new normalizing chromosome(s) result in different cut-offs.
In certain embodiments, a different mask is applied for chromosome Y.
An example of a suitable chromosome Y mask is described in US Provisional Patent ation No. 61/836,057, filed June 17, 2013 [attorney docket no. ARTEP008P], which is incorporated herein by reference for all purposes.
After the system computationally masks the bins, it computationally normalizes the coverage values in the bins that are not excluded by the masks. See block 315. In certain embodiments, the system normalizes the test sample coverage values in each bin (e.g., NES counts per bin) t most or all of the coverage in reference genome or a portion thereof (e.g., the coverage in the robust chromosomes of the nce genome). In some cases, the system normalizes the test sample coverage values (per bin) by dividing the count for the bin under consideration by the total number of all non-excluded sites aligning to all robust chromosomes in the nce genome. In some embodiments, the system normalizes the test sample coverage values (per bin) by performing a linear regression. For instance, the system first calculates coverages for a subset ofbins in robust chromosomes as Ya =intercept + slope * gwpa, where Ya is coverage for bin a, and gwpa is the global profile for the same bin. The system then calculates the ized coverages Zb as: Zb = Yb I (intercept+ slope * gwpb) - 1.
As explained above, a robust chromosome is one that is unlikely to be aneuploid. In certain embodiments, the robust somes are all autosomal chromosomes other than chromosomes 13, 18, and 21. In some embodiments, the robust chromosomes are all mal somes other than chromosomes determined to deviate from a normal diploid genome.
A bin's transformed count value or coverage 1s referred to as a "normalized coverage quantity" for further processing. The normalization is performed using information unique to each sample. Typically, no information from a training set is used. Normalization allows coverage quantities from samples having different library sizes (and consequently different s of reads and tags) to be treated on equal footing. Some of the subsequent process operations use coverage ties derived from training samples which may be sequenced from libraries that are larger or smaller than the ies used for a test sample under consideration.
Without ization based on the number of reads aligned to the entire reference genome (or at least the robust chromosomes), treatment using parameters derived from a training set might not be reliable or generalizable in some implementations.
Figure 3B illustrates the coverage across chromosomes 21, 13, and 18 for many samples. Some ofthe s were processed differently from one another.
As a consequence, one can see a wide sample-to-sample variation at any given genomic position. Normalization removes some of the sample-to-sample variation.
The left panel of Figure 3C depicts normalized coverage quantities across an entire genome.
In the ment of Figure 3A, the system removes or s a "global e" from the normalized coverage quantities produced in operation 315.
See block 317. This operation removes systematic biases in the normalized coverage quantities arising from the structure of the genome, the y generation process, and the sequencing process. In addition, this ion is designed to correct for any systematic linear deviation from the expected profile in any given sample.
In some implementations, the global profile removal involves dividing the ized coverage quantity of each bin by a corresponding expected value of each bin. In other embodiments, the global profile removal involves subtracting an expected value of each bin from the normalized coverage quantity of each bin. The expected value may be obtained from a training set of unaffected samples (or unaffected female samples for the X chromosome). Unaffected samples are samples from individuals known not to have an oidy for the chromosome of st. In some implementations, the global e removal involves subtracting the expected value of each bin (obtained from a training set) from the ized coverage quantity of each bin. In some embodiments, the process uses median values of normalized coverage quantities for each bin as ined using the training set. In other words, the median values are the expected values.
[00325] In some ments, the global profile removal is implemented using a linear correction for the dependence of the sample coverage on the global profile.
As indicated, the global profile is an expected value for each bin as ined from the training set (for example the median value for each bin). These embodiments may employ a robust linear model obtained by fitting the test sample's normalized coverage quantities against the global median profile obtained for each bin. In some embodiments, the linear model is ed by regressing the sample's observed normalized coverage quantities against the global median (or other expectation value) profile.
The linear model is based on an assumption that sample coverage quantities have a linear relationship with the global profile values, which linear relationship should hold for both robust chromosomes/regions and a sequence of interest. See Figure 3D. In such case, a regression of the sample normalized coverage quantities on the global e's expected coverage quantities will produce a line having a slope and intercept. In certain ments, the slope and intercept of such line is used to calculate a "predicted" coverage quantity from the global profile value for a bin. In some implementations, a global e correction involves modeling each bin's ized ge quantity by the predicted coverage I0 quantities for the bin. In some implementations, coverages of the test sequence tags are adjusted by: (i) obtaining a mathematical relation between the coverage ofthe test ce tags versus the expected coverage in a plurality of bins in one or more robust chromosomes or regions, and (ii) applying the mathematical relation to bins in the sequence of interest. In some implementations, the coverages in a test sample are corrected for variation using a linear onship between the expected coverage values from unaffected training samples and coverage values for the test sample in robust somes or other robust regions of the genome. The adjustment results in global-profile-corrected coverages. In some cases, the adjustment involves obtaining coverages for a test sample for a subset of bins in robust somes or regions as follows: Ya =intercept+ slope *gwpa where ya is coverage of bin a for the test sample in one or more robust chromosomes or regions, and gwpa is the global profile for bin a for unaffected training s.
The process then computes a global-profile-corrected coverage zb for a sequence or region ofinterest as: Zb =Yb I cept + slope *gwpb) - I where yb is the observed coverage of bin b for the test sample in the sequence of interest (which may reside outside a robust chromosome or region), and gwpb is the global profile for bin b for unaffected training samples. The denominator (intercept + slope * gwpb) is the coverage for bin b that is ted to be observed in cted test samples based on the relationship estimated from robust regions of the genome.
In the case of a sequence of interest harboring a copy number variation, the observed coverage and hence the -profile-corrected coverage value for bin b will deviate significantly from the coverage of an unaffected sample. For example, the corrected coverage zb would be proportional to fetal on in the case of trisomic sample for bins on the affected chromosome. This process normalizes within sample by computing ept and slope on robust chromosomes, and then evaluates how the genomic region of interest deviates from a relationship (as described by the slope and the intercept) that holds for robust chromosomes within the same sample.
The slope and intercept are obtained from a line as shown in Figure 3D. An example of global e removal is depicted in Figure 3C. The left panel shows a high bin-to-bin variation in normalized coverage quantities across many s. The right panel shows the same normalized coverage quantities after global profile removal as described above.
After the system removes or reduces the global profile variations at block 317, it corrects for in-sample GC (guanine-cytosine) content variations. See block 319. Every bin has its own fractional contribution from GC. The fraction is determined by dividing the number of G and C nucleotides in a bin by the total number of nucleotides in a bin (e.g., 100,000). Some bins will have greater GC fractions than others. As shown in Figures 3E and 3F, different samples t different GC biases. These differences and their corrections will be explained further below. Figures 3E-G show global profile corrected, normalized coverage quantity (per bin) as a function of GC fraction (per bin). Surprisingly, different s exhibit different GC dependence. Some s show monotonically decreasing dependence (as in Figure 3E), while others exhibit a comma shaped dependence (as in Figure 3F and 3G). e these profiles may be unique for each , the correction described in this step is med separately and uniquely for each sample.
In some embodiments, the system computationally arranges bins on the basis of GC fraction as illustrated in s 3E-G. It then corrects the global profile corrected, normalized coverage quantity of a bin using information from other bins with similar GC contents. This correction is applied to each ed bin.
In some processes, each bin is corrected for GC content m the following way. The system computationally selects bins having GC fractions similar 2016/067886 to those of a bin under consideration and then determines a correction parameter from information in the selected bins. In some embodiments, those bins having r GC fractions are selected using an arbitrarily defined cut-off value of similarity. In one example, 2% of all bins are selected. These bins are the 2% having GC content bins most similar to the bin under consideration. For example, the 1% of bins having slightly more GC content and 1% having slightly less GC content are selected.
Using the selected bins, the system computationally determines a correction parameter. In one e, the correction parameter is a representative value of the normalized coverage quantities (after global profile removal) in the selected bins. Examples of such representative value include the median or mean of the normalized coverage quantities in the ed bins. The system applies a calculated correction parameter for a bin under consideration to the normalized coverage quantity (after global profile removal) for the bin under consideration. In some implementations, a representative value (e.g., median value) is subtracted from the normalized coverage ty of the bin under consideration. In some embodiments, the median value (or other representative value) of normalized coverage quantities is selected using only the coverage ties for robust autosomal chromosomes (all autosomes other than chromosomes 13, 18, and 21).
In one example using, e.g., lOOkb bins, each bin will have a unique value of GC fraction, and the bins are d into groups based on their GC fraction content. For example, the bins are divided into 50 groups, where group ries correspond to (0, 2, 4, 6, ... , and 100) quantiles of the %GC bution. A median normalized coverage quantity is calculated for each group of bins from the robust autosomes mapping to the same GC group (in the sample), and then the median value is subtracted from the normalized coverage quantities (for all bins across the entire genome in the same GC group). This applies a GC correction estimated from robust somes within any given sample to the potentially affected chromosomes within the same sample. For example, all bins on robust chromosomes having a GC content between 0.338660 and 0.344720 are d together, the median is calculated for this group and is subtracted from the normalized coverage of the bins within this GC range, which bins may be found anywhere on the genome ding chromosomes 13, 18, 21, and X). In certain embodiments, chromosome Y is excluded from this GC correction process.
Figure 3G shows ation of a GC correction using median ized coverage quantities as a correction parameter as just described. The left panel shows the uncorrected coverage quantities versus GC fraction profile. As shown, the profile has a non-linear shape. The right panel shows the corrected coverage ties. Figure 3H shows the normalized coverages for many samples before GC fraction tion (left panel) and after GC fraction correction (right panel). Figure 31 shows the coefficient ofvariation (CV) ofthe normalized coverages for many test samples before GC fraction correction (red) and after GC fraction correction (green), where GC correction leads to substantially smaller variation in I0 normalized ges.
The above process is a relatively simple implementation of the GC tion. Alternative approaches to correcting for GC bias employ a spline or other non-linear fitting technique, which may be applied in the continuous GC space and does not involve binning coverage quantities by GC content. Examples of suitable I 5 techniques include continuous loess correction and smooth spline correction. A fitting on may be derived from bin-by-bin normalized coverage quantity versus GC content for the sample under consideration. The tion for each bin is calculated by ng the GC content for bin under consideration to the fitting function. For instance, the normalized coverage quantity may be adjusted by subtracting the ed ge value of a spline at the GC content ofthe bin under eration. Alternatively, the adjustment may be achieved by division of the ed coverage value according to the spline fit.
After correcting the GC-dependence m operation 3 I9, the system computationally removes outlier bins in sample under consideration -See block 321.
This operation may be referred to as single sample filtering or trimming. Figure 3J shows that even after GC correction, the coverage still has sample-specific variation within small regions. See for example the coverage at position I. I e8 on chromosome I2 where an unexpectedly high deviation from the expected value s. It is possible that this deviation results from a small copy number variation in the material genome. Alternatively, this may be due to technical reasons in sequencing unrelated to copy number variation. Typically, this operation is only applied to the robust chromosomes.
WO 36059 ] As one example, the systems computationally filters any bins having a GC corrected normalized coverage quantity of more than 3 median te deviations from the median of the GC corrected normalized coverage quantity across all bins in the chromosome harboring the bin under consideration for filtering. In one example, the cut-off value is defined as 3 median absolute deviations adjusted to be consistent with the standard deviation, so actually the f is l .4826*median absolute ions from the . In certain embodiments, this operation is applied to all chromosomes in the sample, including both the robust chromosomes and the somes suspected of aneuploidy.
I0 [00337] In certain implementations, an additional ion which may be characterized as quality control is performed. See block 323. In some embodiments, a quality l metric involves detection of whether any potential denominator chromosomes i.e. "normalizing chromosomes" or "robust chromosomes" are aneuploid or otherwise inappropriate for determining whether the test sample has a copy number variation in a sequence of interest. When the process determines that a robust chromosome is inappropriate, the process may disregard the test sample and make no call. Alternatively, a failure ofthis QC metric may trigger use of an alternate set of normalizing chromosomes for calling. In one example, a quality control method compares actual normalized coverage values for robust chromosomes against ation values for robust autosomal chromosomes. The ation values can be ed by fitting a ariate normal model to the normalized profiles of unaffected training samples, selecting the best model structure according to the likelihood of the data or Bayesian criteria (e.g., the model is selected using Akaike information criterion or possibly Bayesian information criterion), and fixing the best model for use in QC. The normal models ofthe robust chromosomes can be obtained by, for example, using a clustering technique that identifies a probability function having a mean and standard deviation for the chromosome coverages in the normal samples. Of course, other model forms may be used. The process evaluates the likelihood of observed ized coverage in any incoming test sample given the fixed model parameters. It may do this by scoring each incoming test sample with the model to obtain likelihood and thereby identify outliers relative to normal sample set.
Deviation in the likelihood of the test sample from that of the training samples may suggest either an abnormality in izing chromosomes or a sample handling I assay processing artifact that may result in incorrect sample classification. This QC metric can be used to reduce errors in fication associated with either of these sample artifacts. Figure 3K, right panel, shows on the x-axis some number and the y-axis shows normalized some coverage based on a comparison with a QC model obtained as described above. The graphs shows one sample with an excessive coverage for chromosome 2 and other sample with an excessive coverage for chromosome 20. These samples would be eliminated using the QC metric described here or diverted to use an alternate set of izing somes. The left panel ofFigure 3K shows NCV versus likelihood for a chromosome.
[00338] The sequence depicted in Figure 3A may be used for all bins of all chromosomes in the genome. In certain embodiments, a different process is applied to chromosome Y. To calculate chromosome or segment dose, NCV, and/or NSV, the corrected normalized coverage quantities (as determined in Figure 3A) from bins in the chromosomes or segments used in the expressions for dose, NCV, and/or NSV are used. See block 325. In certain embodiments, a mean normalized ge quantity is calculated from all bins in a chromosome of interest, normalizing chromosome, segment of interest, and/or normalizing segment is used to calculate sequence dose, NCV, and/or NSV as described elsewhere herein.
In certain ments, chromosome Y is treated differently. It may be filtered by masking a set of bins unique to the Y chromosome. In some embodiments, the Y chromosome filter is determined according the process in US Provisional Patent Application No. 61/836,057, previously incorporated by reference.
In some embodiments, the filter masks bins that are smaller than those in the filter of the other chromosomes. For example, the Y chromosome mask may filter at the 1 kb level, while the other chromosome masks may filter at the 100 kb level. Nevertheless, the Y chromosome may be normalized at the same bin size as the other chromosomes (e.g., 100 kb).
In certain embodiments, the ed Y chromosome is normalized as described above in operation 315 of Figure 3A. However, otherwise, the Y chromosome is not r corrected. Thus, the Y chromosome bins are not subjected to global profile removal. Similarly, the Y chromosome bins are not ted to GC correction or other filtering steps performed fter. This is because when the sample is sed, the process does not know whether the sample is male or female.
A female sample should have no reads aligning to the Y reference chromosome.
Creating a Sequence Mask ] Some embodiments disclosed herein employ a strategy for filtering out (or masking) non-discriminant sequence reads on a sequence of interest using sequence masks, which leads to higher signal and lower noise, relatively to values calculated by conventional methods, in the coverage values used for CNV evaluation.
Such masks can be identified by various ques. In one embodiment, a mask is identified using a technique illustrated in Figures 4A-4B as explained below in further details.
In some implementations, the mask is identified using a training set of representative samples known to have normal copy number of the ce of interest. Masks may be identified using a que that first normalizes the training set samples, then corrects for systematic variation across a range of sequence (e.g., a profile), and then corrects them for GC variability as described below. The normalization and correction are performed on samples from a training set, not test samples. The mask is identified once and then applied to many test samples.
Figure 4A shows a flow chart of a process 400 for creating such a sequence mask, which can be applied to one or more test samples to remove bins on a sequence of interest from consideration in evaluation of copy number. Process 400 illustrated in Figure 4 uses sequence tag coverage based on the number of sequence tags to obtain a sequence mask. r, similar to the ption above regarding process 100 for determining CNV with reference to Figure 1, other variables or parameters, such as size, size ratio, and methylation level, may be used in addition to or d of ge for process 400. In some implementations, one mask is ted for each of two or more ters. Furthermore, coverage and other parameters may be weighted based on the size of the fragments from which tags are derived. For ease of reading, only coverage is referred to in process 400, but one should note that other parameters, such as size, size ratio, and methylation level, count weighted by size, etc. may be used in the place of coverage.
Process 400 starts by providing a training set ing sequence reads from a plurality of unaffected training samples. Block 402. The process then align the sequence reads of the training set to a reference genome comprising the sequence of interest, thereby ing training sequence tags for the training samples. Block 404. In some embodiments, only uniquely aligned non-redundant tags mapped to cluded sites are used for further analysis. The s involves dividing the reference genome into a ity ofbins and determining for each unaffected training sample a coverage of training sequence tags in each bin for each training sample.
Block 406. The process also determines for each bin an expected coverage of the training sequence tags across all training samples. Block 408. In some embodiments, the expected ge of each bin is the median or means across the training samples.
I0 The expected coverages constitutes a global profile. The process then adjust the coverage of the training sequence tags in each bin for each training sample by removing the variation in the global profile, thereby obtaining global-profile-corrected coverages of the training sequence tags in the bins for each training sample. The process then creates a sequence mask comprising ed and masked bins across the reference genome. Each masked bin has a bution characteristic exceeding a masking threshold. The distribution characteristic is provided for the adjusted coverages of the training sequence tags in the bin across training s. In some implementations, the masking threshold may relate to the observed variation in normalized coverage within a bin across training samples. Bins with high coefficients ofvariation or median absolute deviation of normalized coverage across samples may be identified based on an empirical distribution of the respective metrics. In some ative implementations, the masking threshold may relate to the ed variation in normalized coverage within a bin across training samples. Bins with high coefficients of variation or median absolute ion of normalized coverage across samples may be masked based on an empirical distribution ofthe respective metrics.
In some implementations, separate cut-offs for identifying masked bins, i.e., masking thresholds, are defined for the chromosome of st and for all other chromosomes. Further, separate g thresholds may be defined for each chromosome of interest separately, and a single masking threshold for the set of all non-affected chromosomes. As an example, a mask based on a certain masking threshold is d for chromosome 13 and another masking threshold is used to define a mask for the other chromosomes. Non-affected chromosomes may also have their masking thresholds defined per chromosome.
Various masking threshold combinations may be evaluated for each chromosome of interest. The masking threshold combinations e one mask for the bins of the chromosome of interest and a different mask for the bins of all other chromosomes.
[00347] In one ch, a range of values for coefficient of variation (CV) or measure of sample bution cut-offs is defined as percentiles (e.g., 95, 96, 97, 98, 99) of the empirical distribution of bin CV values and these cut-off values are applied to all autosomes excluding chromosomes of interest. Further, a range of percentile cut-off values for CV is defined for the empirical CV distribution and these cut-off values are applied to a some of interest (e.g., chr 21 ). In some embodiments, the chromosomes of interest are the X chromosome and somes 13, 18, and 21.
Of course, other approaches may be considered; for example, a separate optimization may be performed for each chromosome. Together, the ranges to be optimized in parallel (e.g., one range for a chromosome of interest under consideration and r range for all other chromosomes) define a grid of CV cut-off ations. See Figure 4B. Performance of the system on the training set is evaluated across the two cut-offs (one for normalizing chromosomes (or autosomes other than the chromosome of interest) and one for chromosome of interest), and the best performing combination is chosen for final configuration. This combination may be different for each of the chromosomes of interest. In certain embodiments, performance is evaluated on a tion set instead of the training set, namely, cross-validation is used to evaluate performance.
In some embodiments, the performance optimized to determine f ranges is the cient of variation of chromosome doses (based on a tentative selection of izing chromosomes). The process selects the combination of cut- offs that minimize the CV of the chromosome dose (e.g., ratio) ofthe some of interest using a currently a selected normalizing chromosome (or chromosomes). In one ch, the process tests the performance of each combination of cut-offs in the grid as follows: (1) apply the combination of cut-offs to define masks for all chromosomes and apply those masks to filter the tags of a training set; (2) calculate normalized coverages across the training set of unaffected samples by applying the process of Figure 3A to the filtered tags; (3) determine a entative normalized coverage per chromosome by, e.g., summing the bin's normalized coverages for a chromosome under consideration; (4) calculate chromosome doses using the current normalizing chromosomes, and (5) determine the CVs of the chromosome doses. The process may assess the performance ofthe ed filters by applying them to a set of test samples separated from an original portion ofthe training set. That is, the process splits the original training set into training and testing subsets. The training subset is used to define the mask cut-offs as bed above.
In alternative embodiments, instead of defining masks based on CV of coverages, the masks may be defined by a distribution of mapping quality scores from the alignment results across training samples within the bins. A mapping quality score reflects the uniqueness with which a read is mapped to the reference genome. In other words, mapping quality scores quantify the probability that a read is misaligned. A low mapping y score is associated low uniqueness (high probability of gnment). The uniqueness accounts for one or more errors in the read sequence (as generated by the cer). A detailed description of mapping quality scores is presented in Li H, Ruan J, Durbin R. (2008) g short DNA sequencing reads and calling variants using g quality scores. Genome Research 18: 1851-8, which is incorporated herein by reference in its entirety. In some implementation, the mapping quality score herein is referred to as a MapQ score. Figure 4B shows that MapQ score has a strong monotonous correlation with CV of processed ges.
For instance, bins with CV higher than 0.4 almost completely cluster on the left of the plot in Figure 4B, having MapQ scores lower than about 4. Therefore, masking bins with small MapQ can yield a mask quite similar to one defined by masking bins with high CV.
Samples and Sample Processing Samples Samples that are used for determining a CNV, e.g., chromosomal oidies, l aneuploidies, and the like, can include samples taken from any cell, tissue, or organ in which copy number variations for one or more sequences of st are to be ined. Desirably, the samples contain nucleic acids that are that are present in cells and/or nucleic acids that are "cell-free" (e.g., cIDNA).
In some embodiments it is advantageous to obtain cell-free nucleic acids, e.g., cell-free DNA (cIDNA). Cell-free c acids, including cell-free DNA, can be obtained by vanous methods known in the art from biological samples including but not limited to plasma, serum, and urine (see, e.g., Fan et al., Proc Natl Acad Sci 105: 16266-16271 ; Koide et al., Prenatal Diagnosis 25:604-607 ; Chen et al., Nature Med. 2: 1033-1035 [1996]; Lo et al., Lancet 350: 485-487
[1997]; Botezatu et al., Clin Chem. 46: 1078-1084, 2000; and Su et al., J Mol. Diagn. 6: 101-107 ). To separate cell-free DNA from cells in a sample, vanous methods including, but not limited to fractionation, centrifugation (e.g., density gradient centrifugation), DNA-specific precipitation, or high-throughput cell sorting and/or other separation methods can be used. Commercially available kits for manual and ted separation of cIDNA are available (Roche Diagnostics, Indianapolis, IN, Qiagen, Valencia, CA, Macherey-Nagel, Duren, DE). Biological samples comprising cIDNA have been used in assays to determine the presence or absence of chromosomal alities, e.g., trisomy 21, by sequencing assays that can detect chromosomal aneuploidies and/or various polymorphisms.
[00352] In various ments the cIDNA present m the sample can be enriched specifically or non-specifically prior to use (e.g., prior to preparing a sequencing library). Non-specific enrichment of sample DNA refers to the whole genome amplification of the genomic DNA fragments of the sample that can be used to increase the level of the sample DNA prior to preparing a cIDNA sequencing library. Non-specific enrichment can be the selective enrichment of one of the two genomes present in a sample that comprises more than one genome. For example, non-specific enrichment can be selective of the fetal genome in a maternal sample, which can be obtained by known methods to increase the relative proportion of fetal to maternal DNA in a sample. Alternatively, non-specific enrichment can be the non- selective amplification of both s present in the sample. For example, nonspecific amplification can be of fetal and al DNA in a sample comprising a mixture of DNA from the fetal and maternal genomes. Methods for whole genome ication are known in the art. Degenerate ucleotide-primed PCR (DOP), primer extension PCR technique (PEP) and le cement amplification (MDA) are examples of whole genome amplification methods. In some ments, the sample comprising the mixture of cIDNA from different genomes is un-enriched for cIDNA of the genomes present in the mixture. In other embodiments, the sample comprising the mixture of cIDNA from different genomes is non-specifically enriched for any one ofthe genomes present in the sample.
The sample comprising the nucleic acid(s) to which the methods described herein are applied typically comprises a biological sample ("test sample"), e.g., as described above. In some ments, the c acid(s) to be screened for one or more CNVs is ed or isolated by any of a number ofwell-known methods.
Accordingly, in certain embodiments the sample comprises or consists of a purified or isolated polynucleotide, or it can comprise samples such as a tissue sample, a biological fluid sample, a cell sample, and the like. Suitable biological fluid I0 s include, but are not limited to blood, , serum, sweat, tears, sputum, urine, sputum, ear flow, lymph, saliva, cerebrospinal fluid, ravages, bone marrow suspension, vaginal flow, trans-cervical lavage, brain fluid, ascites, milk, secretions of the respiratory, intestinal and genitourinary tracts, amniotic fluid, milk, and horesis samples. In some embodiments, the sample is a sample that is easily obtainable by non-invasive procedures, e.g., blood, plasma, serum, sweat, tears, sputum, urine, sputum, ear flow, saliva or feces. In certain embodiments the sample is a peripheral blood sample, or the plasma and/or serum fractions of a peripheral blood sample. In other embodiments, the ical sample is a swab or smear, a biopsy specimen, or a cell e. In another embodiment, the sample is a e of two or more biological samples, e.g., a biological sample can se two or more of a biological fluid sample, a tissue sample, and a cell culture sample. As used herein, the terms "blood," "plasma" and "serum" sly encompass ons or processed portions thereof. Similarly, where a sample is taken from a biopsy, swab, smear, etc., the "sample" expressly encompasses a processed on or portion derived from the biopsy, swab, smear, etc.
In certain embodiments, s can be obtained from sources, including, but not limited to, samples from different individuals, samples from different developmental stages of the same or different individuals, samples from different diseased individuals (e.g., individuals with cancer or suspected of having a genetic disorder), normal individuals, s obtained at different stages of a disease in an individual, samples obtained from an individual subjected to different treatments for a disease, samples from individuals subjected to different environmental factors, samples from individuals with predisposition to a pathology, samples individuals with exposure to an infectious disease agent (e.g., HIV), and the like.
In one illustrative, but non-limiting embodiment, the sample is a maternal sample that is obtained from a pregnant female, for example a pregnant woman. In this instance, the sample can be analyzed using the s described herein to provide a prenatal diagnosis of potential chromosomal abnormalities in the fetus. The maternal sample can be a tissue sample, a biological fluid sample, or a cell sample. A biological fluid includes, as miting examples, blood, plasma, serum, sweat, tears, sputum, urine, sputum, ear flow, lymph, saliva, cerebrospinal fluid, I0 ravages, bone marrow suspension, vaginal flow, transcervical , brain fluid, ascites, milk, secretions of the respiratory, intestinal and genitourinary tracts, and leukophoresis s.
In another illustrative, but non-limiting embodiment, the maternal sample is a mixture oftwo or more biological samples, e.g., the ical sample can comprise two or more of a biological fluid , a tissue sample, and a cell culture . In some embodiments, the sample is a sample that is easily obtainable by non-invasive procedures, e.g., blood, plasma, serum, sweat, tears, sputum, urine, milk, sputum, ear flow, saliva and feces. In some embodiments, the biological sample is a peripheral blood sample, and/or the plasma and serum ons thereof. In other embodiments, the biological sample is a swab or smear, a biopsy specimen, or a sample of a cell culture. As disclosed above, the terms "blood," "plasma" and "serum" expressly encompass fractions or processed portions thereof. Similarly, where a sample is taken from a biopsy, swab, smear, etc., the "sample" expressly encompasses a processed fraction or portion derived from the biopsy, swab, smear, etc.
In n embodiments s can also be obtained from in vitro cultured tissues, cells, or other cleotide-containing sources. The cultured samples can be taken from sources including, but not limited to, cultures (e.g., tissue or cells) maintained in different media and conditions (e.g., pH, pressure, or temperature), cultures (e.g., tissue or cells) ined for different periods of , cultures (e.g., tissue or cells) treated with ent factors or reagents (e.g., a drug candidate, or a modulator), or cultures of different types of tissue and/or cells.
Methods of isolating c acids from biological sources are well known and will differ depending upon the nature of the . One of skill in the art can readily e nucleic acid(s) from a source as needed for the method described herein. In some instances, it can be advantageous to fragment the nucleic acid molecules in the nucleic acid sample. Fragmentation can be random, or it can be ic, as achieved, for example, using restriction endonuclease digestion. Methods for random fragmentation are well known in the art, and include, for example, d DNAse digestion, alkali treatment and physical shearing. In one embodiment, sample nucleic acids are obtained from as cIDNA, which is not subjected to fragmentation.
Sequencing Library Preparation In one embodiment, the methods described herein can utilize next generation sequencing technologies (NGS), that allow multiple samples to be sequenced individually as genomic molecules (i.e., singleplex sequencing) or as pooled s comprising indexed genomic molecules (e.g., multiplex sequencing) on a single cing run. These methods can generate up to several hundred n reads of DNA sequences. In s embodiments the sequences of genomic nucleic acids, and/or of indexed genomic nucleic acids can be determined using, for example, the Next Generation Sequencing Technologies (NGS) described herein. In various embodiments analysis ofthe massive amount of ce data obtained using NGS can be performed using one or more processors as described herein.
In various embodiments the use of such sequencing technologies does not involve the preparation of sequencing libraries.
However, m certain ments the sequencmg methods contemplated herein involve the preparation of sequencing libraries. In one illustrative approach, sequencing library preparation involves the production of a random collection of adapter-modified DNA fragments (e.g., polynucleotides) that are ready to be sequenced. Sequencing ies of polynucleotides can be prepared from DNA or RNA, including equivalents, s of either DNA or cDNA, for example, DNA or cDNA that is complementary or copy DNA produced from an RNA template, by the action of reverse transcriptase. The cleotides may originate in doublestranded form (e.g., dsDNA such as genomic DNA fragments, cDNA, PCR amplification products, and the like) or, in certain ments, the polynucleotides may originated m single-stranded form (e.g., ssDNA, RNA, etc.) and have been converted to dsDNA form. By way of illustration, in certain embodiments, single stranded mRNA molecules may be copied into double-stranded cDNAs suitable for use in preparing a sequencing library. The precise ce of the primary polynucleotide les is generally not material to the method of library preparation, and may be known or unknown. In one embodiment, the polynucleotide molecules are DNA molecules. More particularly, in certain embodiments, the cleotide molecules represent the entire genetic complement of an organism or substantially the entire genetic complement of an organism, and are genomic DNA molecules (e.g., cellular DNA, cell free DNA (cIDNA), etc.), that typically include both intron sequence and exon sequence (coding sequence), as well as non-coding regulatory sequences such as er and enhancer sequences. In certain embodiments, the primary polynucleotide molecules comprise human genomic DNA molecules, e.g., cIDNA molecules present in eral blood of a pregnant subject.
[00363] Preparation of sequencing libraries for some NGS sequencing platforms is facilitated by the use of polynucleotides comprising a specific range of fragment sizes. Preparation of such libraries typically involves the fragmentation of large cleotides (e.g. cellular genomic DNA) to obtain polynucleotides in the desired size range.
] Fragmentation can be achieved by any of a number of methods known to those of skill in the art. For e, fragmentation can be achieved by mechanical means including, but not limited to nebulization, sonication and hydroshear. However mechanical fragmentation typically cleaves the DNA backbone at C-0, P-0 and C-C bonds resulting in a heterogeneous mix of blunt and 3'- and 5'-overhanging ends with broken C-0, P-0 and/ C-C bonds (see, e.g., Alnemri and Liwack, J Biol. Chem 265: 17323-17333 [1990]; Richards and Boyer, J Mol Biol 11 :327-240 ) which may need to be repaired as they may lack the requisite sphate for the subsequent tic reactions, e.g., ligation of sequencing adaptors, that are required for preparing DNA for cing.
] In contrast, cIDNA, typically exists as fragments ofless than about 300 base pairs and consequently, fragmentation is not typically necessary for generating a sequencing library using cIDNA samples. lly, r cleotides are forcibly fragmented (e.g., fragmented in vitro), or lly exist as fragments, they are converted to blunt-ended DNA having 5'-phosphates and 3'-hydroxyl. Standard protocols, e.g., protocols for sequencing using, for example, the Illumina platform as described elsewhere herein, instruct users to end-repair sample DNA, to purify the end-repaired products prior to dA-tailing, and to purify the dA-tailing products prior to the adaptor-ligating steps of the library preparation.
Various embodiments of methods of sequence library preparation described herein obviate the need to perform one or more of the steps typically I0 mandated by rd protocols to obtain a ed DNA product that can be sequenced by NGS. An abbreviated method (ABB method), a I-step method, and a 2-step method are examples ofmethods for ation of a sequencing library, which can be found in patent application 13/555,037 filed on July 20, 2012, which is incorporated by nce by its entirety.
Marker Nucleic Acids for tracking and verifying sample integrity In various embodiments verification ofthe integrity of the samples and sample tracking can be accomplished by sequencing mixtures of sample genomic nucleic acids, e.g., cIDNA, and accompanying marker nucleic acids that have been introduced into the samples, e.g., prior to sing.
[00369] Marker nucleic acids can be combined with the test sample (e.g., biological source sample) and subjected to processes that e, for example, one or more of the steps of fractionating the biological source sample, e.g., obtaining an essentially cell-free plasma on from a whole blood sample, purifying nucleic acids from a fractionated, e.g., , or unfractionated biological source sample, e.g., a tissue sample, and sequencing. In some embodiments, sequencing comprises preparing a sequencing library. The sequence or combination of sequences of the marker molecules that are combined with a source sample is chosen to be unique to the source sample. In some embodiments, the unique marker molecules in a sample all have the same sequence. In other ments, the unique marker molecules in a sample are a plurality of sequences, e.g., a combination of two, three, four, five, six, seven, eight, nine, ten, fifteen, twenty, or more different sequences.
] In one embodiment, the integrity of a sample can be verified using a plurality of marker nucleic acid molecules having identical sequences. Alternatively, the identity of a sample can be ed using a plurality of marker nucleic acid molecules that have at least two, at least three, at least four, at least five, at least six, at least seven, at least eight, at least nine, at least ten, at least 11, at least 12, at least 13, at least 14, at least 15, at least 16, at least l 7m, at least 18, at least 19, at least 20, at least 25, at least 30, at least 35, at least 40, at least 50, or more different sequences. cation of the integrity of the plurality of biological samples, i.e., two or more biological samples, requires that each of the two or more samples be marked with marker nucleic acids that have sequences that are unique to each of the plurality of test sample that is being marked. For example, a first sample can be marked with a marker nucleic acid having sequence A, and a second sample can be marked with a marker nucleic acid having sequence B. Alternatively, a first sample can be marked with marker nucleic acid molecules all having sequence A, and a second sample can be marked with a mixture of sequences B and C, wherein sequences A, B and C are marker molecules having different sequences.
The marker nucleic acid(s) can be added to the sample at any stage of sample preparation that occurs prior to library preparation (if libraries are to be prepared) and sequencing. In one ment, marker molecules can be combined with an unprocessed source sample. For example, the marker nucleic acid can be ed in a collection tube that is used to collect a blood . Alternatively, the marker nucleic acids can be added to the blood sample following the blood draw. In one embodiment, the marker nucleic acid is added to the vessel that is used to collect a biological fluid sample, e.g., the marker nucleic acid(s) are added to a blood tion tube that is used to collect a blood sample. In another embodiment, the marker nucleic acid(s) are added to a fraction of the ical fluid sample. For example, the marker nucleic acid is added to the plasma and/or serum fraction of a blood sample, e.g., a maternal plasma sample. In yet another embodiment, the marker molecules are added to a purified , e.g., a sample of nucleic acids that have been purified from a biological sample. For example, the marker nucleic acid is added to a sample of purified maternal and fetal cIDNA. rly, the marker nucleic acids can be added to a biopsy specimen prior to processing the en. In some embodiments, the marker nucleic acids can be combined with a carrier that delivers the marker molecules into the cells of the biological sample. Cell-delivery carriers include pH-sensitive and ic liposomes.
In s embodiments, the marker molecules have antigenomic sequences, that are sequences that are absent from the genome ofthe ical source sample. In an exemplary embodiment, the marker molecules that are used to verify the integrity of a human biological source sample have sequences that are absent from the human genome. In an alternative embodiment, the marker molecules have sequences that are absent from the source sample and from any one or more other known genomes. For example, the marker molecules that are used to verify the integrity of a human biological source sample have sequences that are absent from the human genome and from the mouse genome. The alternative allows for verifying the integrity of a test sample that comprises two or more genomes. For example, the ity of a human cell-free DNA sample obtained from a subject affected by a pathogen, e.g., a bacterium, can be verified using marker molecules having sequences that are absent from both the human genome and the genome of the affecting bacterium. Sequences of genomes of numerous pathogens, e.g., bacteria, viruses, yeasts, fungi, protozoa etc., are publicly available on the World Wide Web at lm.nih.gov/genomes. In another embodiment, marker les are nucleic acids that have sequences that are absent from any known genome. The ces of marker molecules can be randomly generated thmically.
In various embodiments the marker molecules can be llyoccurring deoxyribonucleic acids (DNA), ribonucleic acids or artificial nucleic acid analogs (nucleic acid mimics) including peptide nucleic acids (PNA), morpholino nucleic acid, locked nucleic acids, glycol nucleic acids, and threose nucleic acids, which are guished from naturally-occurring DNA or RNA by changes to the backbone of the molecule or DNA mimics that do not have a phosphodiester backbone. The deoxyribonucleic acids can be from naturally-occurring genomes or can be generated in a laboratory h the use of enzymes or by solid phase chemical synthesis. Chemical methods can also be used to generate the DNA mimics that are not found in nature. Derivatives of DNA are that are available in which the phosphodiester linkage has been replaced but in which the deoxyribose is retained include but are not d to DNA mimics having backbones formed by thioformacetal or a carboxamide linkage, which have been shown to be good structural DNA mimics. Other DNA mimics include lino derivatives and the peptide nucleic acids (PNA), which contain an N-(2-aminoethyl)glycine-based pseudopeptide backbone (Ann Rev Biophys Biomol Struct 24:167-183 [1995]). PNA is an extremely good structural mimic of DNA (or of ribonucleic acid [RNA]), and PNA oligomers are able to form very stable duplex structures with Watson-Crick complementary DNA and RNA (or PNA) oligomers, and they can also bind to targets in duplex DNA by helix invasion (Mol Biotechnol 26:233-248 [2004]. Another good structural mimic/analog of DNA analog that can be used as a marker molecule is phosphorothioate DNA in which one of the non-bridging oxygens is ed by a sulfur. This modification reduces the action of endo-and exonucleases2 ing 5' to 3' and 3' to 5' DNA POL I exonuclease, nucleases SI and Pl, , serum nucleases and snake venom phosphodiesterase.
] The length of the marker molecules can be distinct or indistinct from that ofthe sample nucleic acids, i.e., the length ofthe marker molecules can be similar to that of the sample genomic molecules, or it can be greater or smaller than that of the sample genomic molecules. The length of the marker molecules is measured by the number of nucleotide or nucleotide analog bases that constitute the marker molecule. Marker molecules having lengths that differ from those of the sample genomic molecules can be guished from source nucleic acids using tion methods known in the art. For example, differences in the length of the marker and sample nucleic acid molecules can be determined by electrophoretic tion, e.g., capillary electrophoresis. Size differentiation can be advantageous for quantifying and assessing the quality of the marker and sample nucleic acids. Preferably, the marker c acids are shorter than the genomic nucleic acids, and of sufficient length to e them from being mapped to the genome of the sample. For example, as a 30 base human sequence is needed to uniquely map it to a human genome. Accordingly in certain embodiments, marker les used in sequencing bioassays ofhuman samples should be at least 30 bp in length.
The choice of length of the marker molecule is determined primarily by the sequencing technology that is used to verify the integrity of a source sample.
The length of the sample genomic nucleic acids being sequenced can also be considered. For e, some sequencing technologies employ clonal amplification of polynucleotides, which can require that the genomic polynucleotides that are to be clonally amplified be of a minimum . For example, sequencing using the Illumina GAii sequence analyzer includes an in vitro clonal amplification by bridge PCR (also known as cluster amplification) of polynucleotides that have a minimum length of 11 Obp, to which rs are ligated to provide a nucleic acid of at least 200 bp and less than 600 bp that can be clonally amplified and sequenced. In some embodiments, the length of the r-ligated marker molecule is between about 200bp and about 600bp, between about 250bp and 550bp, between about 300bp and 500bp, or between about 350 and 450. In other embodiments, the length of the adaptor-ligated marker molecule is about 200bp. For e, when sequencing fetal cIDNA that is present in a maternal sample, the length ofthe marker molecule can be chosen to be r to that of fetal cIDNA molecules. Thus, in one embodiment, the length of the marker molecule used in an assay that comprises massively parallel sequencing of cIDNA in a maternal sample to determine the presence or absence of a fetal chromosomal aneuploidy, can be about 150 bp, about 160bp, 170 bp, about 180bp, about 190bp or about 200bp; preferably, the marker molecule is about 170 pp.
Other sequencing approaches, e.g., SOLiD sequencing, Polony Sequencing and 454 sequencing use on PCR to clonally y DNA molecules for sequencing, and each technology dictates the minimum and the maximum length of the molecules that are to be amplified. The length of marker molecules to be sequenced as ly amplified nucleic acids can be up to about 600bp. In some embodiments, the length ofmarker molecules to be sequenced can be greater than 600bp.
Single molecule sequencing technologies, that do not employ clonal amplification of molecules, and are capable of sequencing nucleic acids over a very broad range of template lengths, in most situations do not require that the molecules to be sequenced be of any specific length. However, the yield of sequences per unit mass is dependent on the number of 3' end hydroxyl groups, and thus having vely short templates for sequencing is more efficient than having long templates.
Ifstarting with nucleic acids longer than 1000 nt, it is generally advisable to shear the nucleic acids to an average length of 100 to 200 nt so that more sequence information can be ted from the same mass of c acids. Thus, the length ofthe marker molecule can range from tens of bases to thousands of bases. The length of marker molecules used for single molecule sequencing can be up to about 25bp, up to about 50bp, up to about 75bp, up to about lOObp, up to about 200bp, up to about 300bp, up to about 400bp, up to about 500bp, up to about 600bp, up to about 700bp, up to about 800 bp, up to about 900bp, up to about 1OOObp, or more in length.
The length chosen for a marker molecule is also ined by the length of the genomic nucleic acid that is being sequenced. For example, cIDNA circulates in the human bloodstream as c fragments of cellular genomic DNA Fetal cIDNA molecules found in the plasma of pregnant women are generally r than maternal cIDNA molecules (Chan et al., Clin Chem 50:8892 [2004]). Size fractionation of circulating fetal DNA has confirmed that the average length of circulating fetal DNA fragments is <300 bp, while maternal DNA has been estimated to be between about 0.5 and 1 Kb (Li et al., Clin Chem, 50: 1002-1011 [2004]).
These findings are consistent with those ofFan et al., who determined using NGS that fetal cIDNA is rarely >340bp (Fan et al., Clin Chem 56: 1279-1286 [2010]). DNA isolated from urine with a standard silica-based method consists oftwo fractions, high lar weight DNA, which ates from shed cells and low molecular weight (150-250 base pair) on of transrenal DNA A) (Botezatu et al., Clin Chem. 46: 1078-1084, 2000; and Su et al., J Mol. Diagn. 6: 101-107, 2004). The application of newly developed technique for isolation of cell-free c acids from body fluids to the isolation of transrenal nucleic acids has revealed the presence in urine of DNA and RNA fragments much shorter than 150 base pairs (U.S. Patent Application Publication No. 20080139801). In embodiments, wherein cIDNA is the genomic nucleic acid that is sequenced, marker molecules that are chosen can be up to about the length of the cIDNA For example, the length of marker molecules used in maternal cIDNA s to be sequenced as single nucleic acid molecules or as clonally amplified nucleic acids can be between about 100 bp and 600. In other embodiments, the sample genomic nucleic acids are fragments of larger molecules.
For example, a sample c nucleic acid that is sequenced is fragmented cellular DNA In embodiments, when fragmented cellular DNA is sequenced, the length of the marker molecules can be up to the length of the DNA fragments. In some embodiments, the length of the marker molecules is at least the minimum length required for mapping the sequence read uniquely to the riate reference genome.
In other embodiments, the length of the marker molecule is the minimum length that is required to exclude the marker molecule from being mapped to the sample reference genome.
In addition, marker molecules can be used to verify samples that are not assayed by nucleic acid sequencing, and that can be verified by common biotechniques other than sequencing, e.g., real-time PCR.
Sample controls (e.g.• in process positive controls for sequencing and/or analysis).
In s embodiments marker sequences uced into the samples, e.g., as described above, can function as positive controls to verify the accuracy and efficacy of sequencing and subsequent processing and analysis.
] Accordingly, compositions and method for providing an ess positive control (IPC) for sequencing DNA in a sample are provided. In certain embodiments, ve controls are provided for sequencing cIDNA in a sample comprising a mixture ofgenomes are provided. An IPC can be used to relate baseline shifts in sequence information obtained from different sets of samples, e.g., samples that are sequenced at different times on ent sequencing runs. Thus, for example, an IPC can relate the sequence information obtained for a maternal test sample to the sequence ation ed from a set of qualified samples that were sequenced at a different time.
Similarly, m the case of segment analysis, an IPC can relate the sequence information obtained from a subject for particular segment(s) to the sequence obtained from a set of qualified samples (of r sequences) that were sequenced at a different time. In certain embodiments an IPC can relate the sequence information obtained from a subject for particular cancer-related loci to the sequence information obtained from a set of qualified samples (e.g., from a known amplification/deletion, and the like).
[00382] In addition, IPCs can be used as s to track sample(s) through the sequencing process. IPCs can also provide a qualitative positive ce dose value, e.g., NCV, for one or more aneuploidies of chromosomes of interest, e.g., y 21, trisomy 13, trisomy 18 to provide proper interpretation, and to ensure the ability and accuracy of the data. In certain embodiments IPCs can be created to comprise nucleic acids from male and female genomes to provide doses for chromosomes X and Y in a maternal sample to determine whether the fetus is male.
The type and the number of cess controls depends on the type or nature of the test needed. For example, for a test requiring the sequencing of DNA from a sample comprising a mixture ofgenomes to determine whether a chromosomal aneuploidy exists, the in-process control can comprise DNA obtained from a sample known comprising the same chromosomal aneuploidy that is being tested. In some embodiments, the IPC includes DNA from a sample known to comprise an aneuploidy of a chromosome of interest. For example, the IPC for a test to determine the presence or absence of a fetal trisomy, e.g., trisomy 21, in a maternal sample comprises DNA obtained from an individual with trisomy 21. In some embodiments, the IPC comprises a mixture of DNA obtained from two or more individuals with different aneuploidies. For example, for a test to determine the presence or absence of trisomy 13, y 18, trisomy 21, and monosomy X, the IPC ses a combination of DNA samples obtained from pregnant women each carrying a fetus with one of the trisomies being . In addition to complete chromosomal aneuploidies, IPCs can be created to e ve controls for tests to determine the presence or absence of partial aneuploidies.
An IPC that serves as the l for detecting a single aneuploidy can be created using a mixture of cellular genomic DNA obtained from a two subjects one being the contributor ofthe oid genome. For example, an IPC that is created as a control for a test to determine a fetal trisomy, e.g., trisomy 21, can be created by combining genomic DNA from a male or female subject carrying the trisomic chromosome with c DNA with a female subject known not to carry the trisomic chromosome. Genomic DNA can be extracted from cells of both subjects, and sheared to provide fragments of n about 100 - 400 bp, between about 150- 350 bp, or between about 200-300 bp to simulate the circulating cIDNA fragments in maternal samples. The proportion of fragmented DNA from the subject carrying the aneuploidy, e.g., trisomy 21, is chosen to te the proportion of ating fetal cIDNA found in maternal samples to provide an IPC comprising a mixture of fragmented DNA comprising about 5%, about 10%, about 15%, about 20%, about 25%, about 30%, of DNA from the t carrying the oidy. The IPC can comprise DNA from different subjects each carrying a different aneuploidy. For example, the IPC can comprise about 80% of the unaffected female DNA, and the remaining 20% can be DNA from three different subjects each carrying a trisomic chromosome 21, a ic chromosome 13, and a trisomic chromosome 18. The mixture of fragmented DNA is prepared for cing. Processing ofthe mixture of fragmented DNA can comprise preparing a sequencing library, which can be sequenced using any massively parallel methods in singleplex or multiplex fashion.
Stock solutions of the genomic IPC can be stored and used in multiple stic tests.
Alternatively the IPC can be d using cIDNA obtained from a mother known to carry a fetus with a known chromosomal oidy. For example, cIDNA can be obtained from a pregnant woman carrying a fetus with trisomy 21.
The cIDNA is extracted from the maternal sample, and cloned into a bacterial vector and grown in bacteria to provide an ongoing source of the IPC. The DNA can be extracted from the bacterial vector using restriction enzymes. atively, the cloned cIDNA can be amplified by, e.g., PCR. The IPC DNA can be processed for sequencing in the same runs as the cIDNA from the test samples that are to be analyzed for the presence or absence of chromosomal aneuploidies.
While the creation ofIPCs is described above with respect to trisomies, it will be appreciated that IPCs can be created to reflect other partial aneuploidies including for example, various segment amplification and/or deletions. Thus, for example, where various cancers are known to be associated with particular amplifications (e.g., breast cancer associated with 20Q13) IPCs can be created that incorporate those known ications.
Sequencing Methods As indicated above, the prepared s (e.g., Sequencing Libraries) are sequenced as part ofthe ure for identifying copy number variation(s). Any of a number of sequencing technologies can be utilized.
] Some sequencing technologies are available commercially, such as the sequencing-by-hybridization platform from Affymetrix Inc. (Sunnyvale, CA) and the cing-by-synthesis platforms from 454 Life Sciences (Bradford, CT), Illumina/Solexa rd, CA) and Helicos Biosciences (Cambridge, MA), and the sequencing-by-ligation platform from Applied Biosystems (Foster City, CA), as described below. In on to the single molecule sequencing performed using sequencing-by-synthesis of Helicos Biosciences, other single molecule sequencing technologies include, but are not limited to, the SMRT™ technology of Pacific ences, the ION TORRENT1M technology, and re sequencing ped for example, by Oxford Nanopore Technologies.
While the automated Sanger method 1s considered as a 'first generation' technology, Sanger sequencing including the ted Sanger sequencmg, can also be employed in the methods described herein. Additional suitable sequencing methods include, but are not limited to nucleic acid imaging technologies, e.g., atomic force microscopy (AFM) or ission on copy (TEM). Illustrative sequencing technologies are described in greater detail below.
In one illustrative, but non-limiting, embodiment, the methods described herein comprise obtaining sequence ation for the nucleic acids in a test , e.g., cIDNA in a maternal sample, cIDNA or ar DNA in a subject being screened for a cancer, and the like, using Illumina's sequencing-by-synthesis and reversible terminator-based sequencing chemistry (e.g. as described in Bentley et al., Nature 6:53-59 [2009]). Template DNA can be genomic DNA, e.g., cellular DNA or cIDNA. In some embodiments, genomic DNA from isolated cells is used as the te, and it is fragmented into lengths of several hundred base pairs. In other ments, cIDNA is used as the template, and fragmentation is not required as cIDNA exists as short fragments. For example fetal cIDNA circulates in the bloodstream as nts approximately 170 base pairs (bp) in length (Fan et al., Clin Chem 56: 1279-1286 [2010]), and no fragmentation of the DNA is required prior to sequencing. Illumina's sequencing technology relies on the attachment of fragmented genomic DNA to a planar, optically transparent surface on which oligonucleotide anchors are bound. Template DNA is end-repaired to generate 5'-phosphorylated blunt ends, and the polymerase activity of Klenow fragment is used to add a single A base to the 3' end of the blunt phosphorylated DNA fragments. This addition prepares the DNA fragments for ligation to oligonucleotide adapters, which have an overhang of a single T base at their 3' end to increase ligation efficiency. The adapter oligonucleotides are complementary to the flow-cell anchor oligos (not to be confused with the anchor/anchored reads in the analysis of repeat expansion). Under limitingdilution conditions, adapter-modified, single-stranded template DNA is added to the flow cell and immobilized by ization to the anchor oligos. Attached DNA fragments are ed and bridge amplified to create an ultra-high density sequencing flow cell with hundreds of millions of clusters, each containing about 1,000 copies of the same template. In one embodiment, the randomly nted genomic DNA is amplified using PCR before it is subjected to cluster amplification.
Alternatively, an amplification-free (e.g., PCR free) genomic library preparation is used, and the randomly fragmented genomic DNA is enriched using the cluster amplification alone ewa et al., Nature Methods 6:291-295 [2009]). The templates are sequenced using a robust olor DNA sequencing-by-synthesis technology that employs reversible terminators with removable fluorescent dyes.
High-sensitivity fluorescence detection is achieved using laser excitation and total internal reflection optics. Short sequence reads of about tens to a few hundred base pairs are d against a reference genome and unique mapping of the short ce reads to the reference genome are identified using specially developed data analysis pipeline software. After completion of the first read, the templates can be regenerated in situ to enable a second read from the opposite end of the fragments.
Thus, either single-end or paired end sequencing ofthe DNA fragments can be used. s embodiments of the disclosure may use sequencing by synthesis that allows paired end sequencing. In some embodiments, the sequencing by synthesis platform by Illumina involves clustering fragments. Clustering is a process in which each fragment molecule is isothermally amplified. In some ments, as the example described here, the nt has two different adaptors attached to the two ends of the fragment, the adaptors allowing the fragment to hybridize with the two different oligos on the surface of a flow cell lane. The fragment further includes or is connected to two index sequences at two ends of the fragment, which index sequences provide labels to identify different s in multiplex sequencing. In some sequencing rms, a fragment to be sequenced is also referred to as an insert.
In some implementation, a flow cell for clustering in the Illumina platform is a glass slide with lanes. Each lane is a glass l coated with a lawn of two types of oligos. Hybridization is enabled by the first ofthe two types of oligos on the surface. This oligo is complementary to a first adapter on one end ofthe nt.
A polymerase creates a compliment strand of the hybridized fragment. The doublestranded molecule is denatured, and the original template strand is washed away. The remaining strand, in parallel with many other remaining strands, is clonally amplified through bridge application.
In bridge amplification, a strand folds over, and a second adapter region on a second end of the strand hybridizes with the second type of oligos on the flow cell surface. A polymerase generates a complimentary strand, forming a stranded bridge le. This double-stranded molecule is denatured resulting in two single-stranded molecules tethered to the flow cell through two different oligos.
The process is then repeated over and over, and occurs simultaneously for millions of clusters resulting in clonal amplification of all the fragments. After bridge amplification, the e strands are cleaved and washed off, leaving only the forward strands. The 3' ends are blocked to prevent unwanted priming.
After clustering, sequencing starts with extending a first sequencing primer to generate the first read. With each cycle, fluorescently tagged nucleotides compete for addition to the growing chain. Only one is incorporated based on the sequence of the template. After the addition of each nucleotide, the r is excited by a light source, and a characteristic fluorescent signal is emitted. The number of cycles determines the length of the read. The emission wavelength and the signal ity determine the base call. For a given r all identical strands are read simultaneously. Hundreds of millions of rs are sequenced in a massively el manner. At the tion of the first read, the read product is washed away.
In the next step of protocols involving two index primers, an index 1 primer is introduced and hybridized to an index 1 region on the template. Index regions provide identification of fragments, which is useful for de-multiplexing samples in a multiplex sequencing process. The index 1 read is ted similar to the first read. After completion of the index 1 read, the read product is washed away and the 3' end of the strand is de-protected. The template strand then folds over and binds to a second oligo on the flow cell. An index 2 sequence is read in the same manner as index 1. Then an index 2 read product is washed off at the completion of the step.
[00396] After reading two s, read 2 initiates by using polymerases to extend the second flow cell oligos, forming a double-stranded bridge. This doublestranded DNA is red, and the 3' end is blocked. The original forward strand is cleaved off and washed away, leaving the reverse strand. Read 2 begins with the introduction of a read 2 sequencing primer. As with read 1, the sequencing steps are repeated until the desired length is achieved. The read 2 product is washed away.
This entire s generates millions of reads, representing all the fragments.
Sequences from pooled sample ies are separated based on the unique indices introduced during sample preparation. For each sample, reads of similar stretches of base calls are locally clustered. Forward and reversed reads are paired creating uous sequences. These contiguous sequences are aligned to the reference genome for variant identification.
[00397] The sequencing by synthesis example bed above involves paired end reads, which is used in many of the embodiments of the disclosed methods.
Paired end sequencing involves 2 reads from the two ends of a nt. When a pair of reads are mapped to a reference sequence, the base-pair ce between the two reads can be determined, which distance can then be used to determine the length of the fragments from which the reads were obtained. In some instances, a fragment straddling two bins would have one of its pair-end read aligned to one bin, and another to an adjacent bin. This gets rarer as the bins get longer or the reads get shorter. s methods may be used to account for the bin-membership of these fragments. For instance, they can be omitted in determining fragment size ncy of a bin; they can be counted for both ofthe adjacent bins; they can be assigned to the bin that encompasses the larger number of base pairs of the two bins; or they can be assigned to both bins with a weight related to portion ofbase pairs in each bin.
Paired end reads may use insert of different length (i.e., different fragment size to be sequenced). As the default meaning in this sure, paired end reads are used to refer to reads obtained from s insert lengths. In some ces, to distinguish short-insert paired end reads from nserts paired end reads, the latter is also referred to as mate pair reads. In some embodiments involving mate pair reads, two biotin junction adaptors first are ed to two ends of a relatively long insert (e.g., several kb). The biotin junction adaptors then link the two ends of the insert to form a circularized molecule. A sub-fragment encompassing the biotin junction adaptors can then be obtained by further fragmenting the circularized molecule. The sub-fragment including the two ends of the original fragment in opposite sequence order can then be sequenced by the same procedure as for short- insert paired end sequencing described above. Further details of mate pair sequencing using an na platform is shown in an online publication at the following URL, which 1s incorporated by reference by its entirety: res I. lilluminal. ocuments/products/technotes/technote_nextera_matepair_data_pr ocessing. Additional information about paired end sequencing can be found in US Patent No. 7601499 and US Patent Publication No. 2012/0,053,063, which are incorporated by reference with regard to als on paired end sequencing s and apparatuses.
After sequencing of DNA fragments, sequence reads of predetermined length, e.g., 100 bp, are mapped or aligned to a known reference genome. The mapped or aligned reads and their corresponding locations on the reference ce are also referred to as tags. In one embodiment, the reference genome sequence is the NCBI36/hgl8 sequence, which is ble on the world wide web at genome dot ucsc dot edu/cgi-bin/hgGateway?org=Human&db=hgl 8&hgsid=166260105). atively, the nce genome sequence is the GRCh37/hgl9, which is available on the world wide web at genome dot ucsc dot i-bin/hgGateway. Other sources of public sequence information include GenBank, dbEST, dbSTS, EMBL (the European Molecular Biology Laboratory), and the DDBJ (the DNA Databank of Japan). A number of computer algorithms are available for aligning sequences, including without limitation BLAST (Altschul et al., 1990), BLITZ (MPsrch) (Sturrock & Collins, 1993), FASTA (Person & Lipman, 1988), BOWTIE ead et al., Genome Biology 10:R25. l-R25.10 [2009]), or ELAND (Illumina, Inc., San Diego, CA, USA). In one embodiment, one end of the clonally expanded copies of the plasma cIDNA molecules is ced and processed by bioinformatics alignment analysis for the na Genome Analyzer, which uses the ent Large-Scale Alignment ofNucleotide Databases (ELAND) software.
In one illustrative, but non-limiting, embodiment, the methods described herein comprise obtaining sequence information for the nucleic acids in a test sample, e.g., cIDNA in a maternal sample, cIDNA or cellular DNA in a subject being screened for a , and the like, using single molecule sequencing technology of the Helicos True Single Molecule Sequencing (tSMS) technology (e.g. as described in Harris T.D. et al., Science 320: 106-109 [2008]). In the tSMS technique, a DNA sample is cleaved into strands of approximately 100 to 200 nucleotides, and a polyA sequence is added to the 3' end of each DNA strand. Each strand is labeled by the addition of a fluorescently labeled adenosine tide. The DNA s are then hybridized to a flow cell, which contains millions of T capture sites that are immobilized to the flow cell e. In certain embodiments the templates can be at a density of about 100 million templates/cm2. The flow cell is then loaded into an ment, e.g., HeliScope™ sequencer, and a laser illuminates the surface of the flow cell, revealing the position of each template. A CCD camera can map the position of the templates on the flow cell e. The template scent label is then cleaved and washed away. The sequencing reaction begins by introducing a DNA polymerase and a fluorescently labeled nucleotide. The oligo- T nucleic acid serves as a pnmer. The polymerase incorporates the labeled nucleotides to the primer in a template directed manner. The polymerase and unincorporated tides are removed. The tes that have directed incorporation of the fluorescently labeled nucleotide are discerned by imaging the flow cell surface. After imaging, a cleavage step removes the fluorescent label, and the process is repeated with other fluorescently labeled nucleotides until the desired read length is achieved. Sequence information is collected with each nucleotide addition step. Whole genome sequencing by single molecule sequencing technologies excludes or typically obviates PCR-based ication in the preparation of the sequencing libraries, and the methods allow for direct measurement of the sample, rather than measurement of copies ofthat sample.
In another rative, but miting embodiment, the methods described herein comprise obtaining sequence information for the nucleic acids in the test sample, e.g., cIDNA in a maternal test , cIDNA or cellular DNA in a subject being screened for a cancer, and the like, using the 454 sequencing (Roche) (e.g. as described in Margulies, M. et al. Nature 437:376-380 [2005]). 454 sequencing typically involves two steps. In the first step, DNA is sheared into fragments of approximately 300-800 base pairs, and the fragments are blunt-ended.
Oligonucleotide adaptors are then ligated to the ends of the fragments. The adaptors serve as primers for amplification and sequencing of the fragments. The fragments can be attached to DNA capture beads, e.g., streptavidin-coated beads using, e.g., Adaptor B, which contains 5'-biotin tag. The fragments attached to the beads are PCR amplified within droplets of an ter emulsion. The result is multiple copies of ly amplified DNA fragments on each bead. In the second step, the beads are captured in wells (e.g., picoliter-sized wells). Pyrosequencing is med on each DNA nt in parallel. Addition of one or more tides generates a light signal that is recorded by a CCD camera in a sequencing ment. The signal strength is proportional to the number of nucleotides incorporated. Pyrosequencing makes use of pyrophosphate (PPi) which is released upon tide addition. PPi is converted to ATP by ATP sulfurylase in the presence of adenosine 5' phosphosulfate.
Luciferase uses ATP to convert luciferin to oxyluciferin, and this reaction generates light that is ed and analyzed.
[00402] In another illustrative, but non-limiting, embodiment, the methods described herein comprises obtaining sequence information for the c acids in the test sample , e.g., cIDNA in a maternal test sample, cIDNA or cellular DNA in a subject being screened for a cancer, and the like, using the SOLiD™ technology (Applied Biosystems). In SOLiD™ sequencing-by-ligation, genomic DNA is sheared into fragments, and adaptors are ed to the 5' and 3' ends of the fragments to generate a fragment library. Alternatively, internal adaptors can be introduced by ligating rs to the 5' and 3' ends of the fragments, circularizing the nts, digesting the circularized fragment to generate an internal adaptor, and ing adaptors to the 5' and 3' ends of the resulting fragments to generate a mate-paired library. Next, clonal bead populations are ed in microreactors containing beads, primers, template, and PCR components. Following PCR, the templates are denatured and beads are enriched to separate the beads with extended templates.
Templates on the selected beads are subjected to a 3' modification that permits bonding to a glass slide. The sequence can be determined by sequential hybridization and ligation of lly random oligonucleotides with a central determined base (or pair of bases) that is identified by a specific fluorophore. After a color is recorded, the ligated oligonucleotide is cleaved and removed and the process is then repeated.
In another illustrative, but non-limiting, embodiment, the methods described herein comprise obtaining sequence ation for the nucleic acids in the test sample, e.g., cIDNA in a maternal test sample, cIDNA or cellular DNA in a subject being screened for a cancer, and the like, using the single molecule, real-time ) sequencing technology of Pacific Biosciences. In SMRT sequencing, the continuous incorporation of dye-labeled nucleotides is imaged during DNA synthesis.
Single DNA polymerase molecules are attached to the bottom surface of individual ode ngth detectors (ZMW detectors) that obtain sequence information while phospholinked nucleotides are being incorporated into the growing primer strand. A ZMW detector comprises a confinement structure that enables observation of incorporation of a single nucleotide by DNA polymerase against a background of fluorescent nucleotides that rapidly diffuse in an out of the ZMW (e.g., in microseconds). It typically takes several milliseconds to incorporate a nucleotide into a growing strand. During this time, the fluorescent label is excited and produces a fluorescent signal, and the scent tag is cleaved off Measurement of the corresponding fluorescence of the dye indicates which base was incorporated. The process is ed to provide a sequence.
In another illustrative, but non-limiting embodiment, the methods described herein comprise ing sequence ation for the nucleic acids in the test sample, e.g., cIDNA in a al test sample, cIDNA or cellular DNA in a subject being screened for a cancer, and the like, using nanopore cing (e.g. as described in Soni GV and Meller A Clin Chem 53: 1996-2001 [2007]). re sequencing DNA analysis techniques are developed by a number of companies, including, for example, Oxford Nanopore Technologies (Oxford, United Kingdom), Sequenom, NABsys, and the like. Nanopore sequencing is a single-molecule sequencing technology whereby a single molecule of DNA is sequenced directly as it passes h a nanopore. A nanopore is a small hole, typically of the order of 1 ter in diameter. Immersion of a nanopore in a conducting fluid and application of a potential (voltage) across it results in a slight electrical current due to conduction of ions through the nanopore. The amount of current that flows is sensitive to the size and shape ofthe nanopore. As a DNA molecule passes through a nanopore, each nucleotide on the DNA molecule obstructs the nanopore to a different degree, ng the magnitude of the current through the nanopore in different degrees. Thus, this change in the current as the DNA molecule passes through the nanopore provides a read ofthe DNA sequence.
[00405] In another illustrative, but non-limiting, embodiment, the methods bed herein ses obtaining sequence information for the nucleic acids in the test sample, e.g., cIDNA in a al test sample, cIDNA or cellular DNA in a subject being screened for a cancer, and the like, using the chemical-sensitive field WO 36059 effect transistor (chemFET) array (e.g., as described in U.S. Patent Application ation No. 2009/0026082). In one example of this technique, DNA molecules can be placed into on chambers, and the template molecules can be ized to a cing primer bound to a polymerase. Incorporation of one or more triphosphates into a new nucleic acid strand at the 3' end ofthe cing primer can be discerned as a change in current by a chemFET. An array can have multiple chemFET sensors. In another example, single nucleic acids can be attached to beads, and the nucleic acids can be amplified on the bead, and the individual beads can be erred to individual reaction chambers on a chemFET array, with each chamber having a chemFET sensor, and the nucleic acids can be sequenced.
In another embodiment, the present method comprises obtaining sequence information for the nucleic acids in the test sample, e.g., cIDNA in a maternal test sample, using transmission electron microscopy (TEM). The method, termed Individual Molecule Placement Rapid Nano Transfer (IMPRNT), comprises utilizing single atom resolution ission electron microscope imaging of high- molecular weight (150kb or greater) DNA selectively labeled with heavy atom markers and arranging these molecules on ultra-thin films in ultra-dense (3nm strandto-strand ) parallel arrays with consistent base-to-base spacing. The electron microscope is used to image the molecules on the films to determine the position of the heavy atom markers and to extract base sequence information from the DNA. The method is further described in PCT patent publication . The method allows for cing complete human genomes in less than ten minutes.
In another embodiment, the DNA sequencing technology is the Ion Torrent single molecule cing, which pairs semiconductor technology with a simple sequencing try to directly translate chemically encoded information (A, C, G, T) into digital information (0, 1) on a nductor chip. In nature, when a nucleotide is orated into a strand of DNA by a polymerase, a hydrogen ion is released as a byproduct. Ion Torrent uses a high-density array of micro-machined wells to perform this biochemical process in a ely parallel way. Each well holds a different DNA molecule. Beneath the wells is an ion-sensitive layer and beneath that an ion sensor. When a nucleotide, for example a C, is added to a DNA template and is then orated into a strand of DNA, a hydrogen ion will be released. The charge from that ion will change the pH of the solution, which can be WO 36059 detected by Ion Torrent'sion sensor. The sequencer-essentially the world's smallest solid-state pH meter-calls the base, going directly from chemical information to digital information. The Ion personal Genome Machine (PGM™) sequencer then sequentially floods the chip with one nucleotide after another. If the next nucleotide that floods the chip is not a match. No voltage change will be recorded and no base will be called. Ifthere are two identical bases on the DNA strand, the voltage will be double, and the chip will record two identical bases called. Direct ion allows recordation ofnucleotide incorporation in seconds.
In another ment, the present method compnses obtaining I0 sequence information for the nucleic acids in the test sample, e.g., cIDNA in a maternal test sample, using sequencing by hybridization. Sequencing-byhybridization comprises contacting the plurality of polynucleotide sequences with a plurality of polynucleotide probes, wherein each of the plurality of cleotide probes can be optionally tethered to a substrate. The ate might be flat surface sing an array of known nucleotide sequences. The pattern of hybridization to the array can be used to determine the polynucleotide sequences present in the . In other embodiments, each probe is tethered to a bead, e.g., a magnetic bead or the like. Hybridization to the beads can be ined and used to identify the plurality of polynucleotide sequences within the sample.
[00409] In some ments of the methods described , the mapped sequence tags comprise sequence reads of about 20bp, about 25bp, about 30bp, about 35bp,about40bp,about45bp,about50bp,about55bp,about60bp, about65bp, about 70bp, about 75bp, about 80bp, about 85bp, about90bp, about 95bp, about lOObp, about llObp, about 120bp, about 130, about 140bp, about 150bp, about 200bp, about 250bp, about 300bp, about 350bp, about 400bp, about 450bp, or about 500bp. It is expected that technological advances will enable single-end reads of r than 500bp enabling for reads of greater than about IOOObp when paired end reads are generated. In one embodiment, the mapped sequence tags comprise sequence reads that are 36bp. g of the sequence tags is achieved by comparing the sequence of the tag with the sequence of the reference to determine the chromosomal origin of the sequenced nucleic acid (e.g. cIDNA) molecule, and specific genetic sequence information is not needed. A small degree ofmismatch (0-2 mismatches per sequence tag) may be allowed to account for minor polymorphisms that may exist n the reference genome and the genomes in the mixed sample.
] A plurality of sequence tags are typically obtained per sample. In some embodiments, at least about 3 x 106 sequence tags, at least about 5 x 106 sequence tags, at least about 8 x 106 sequence tags, at least about 10 x 106 sequence tags, at least about 15 x 106 sequence tags, at least about 20 x 106 sequence tags, at least about 30 x 106 sequence tags, at least about 40 x 106 sequence tags, or at least about 50 x 106 sequence tags comprising between 20 and 40bp reads, e.g., 36bp, are obtained from mapping the reads to the reference genome per sample. In one embodiment, all the sequence reads are mapped to all regions of the reference genome. In one embodiment, the tags that have been mapped to all regions, e.g., all chromosomes, of the reference genome are counted, and the CNV, i.e., the over- or under-representation of a sequence of interest, e.g., a chromosome or portion thereof, in the mixed DNA sample is determined. The method does not require entiation between the two genomes.
] The accuracy required for correctly determining whether a CNV, e.g., aneuploidy, is present or absent in a sample, is predicated on the variation of the number of sequence tags that map to the reference genome among samples within a sequencing run (inter-chromosomal variability), and the variation of the number of sequence tags that map to the reference genome in ent sequencing runs (intersequencing variability). For example, the variations can be particularly pronounced for tags that map to GC-rich or GC-poor reference sequences. Other variations can result from using ent protocols for the extraction and purification of the nucleic acids, the preparation of the sequencing libraries, and the use of different sequencing platforms. The present method uses sequence doses osome doses, or segment doses) based on the dge of normalizing sequences (normalizing chromosome sequences or normalizing segment sequences), to intrinsically account for the accrued variability stemming from interchromosomal -run), and inter-sequencing (interrun ) and rm-dependent ility. Chromosome doses are based on the knowledge of a normalizing chromosome sequence, which can be composed of a single chromosome, or of two or more chromosomes selected from chromosomes 1- 22, X, and Y. Alternatively, normalizing chromosome sequences can be ed of a single chromosome segment, or of two or more segments of one chromosome or of two or more chromosomes. Segment doses are based on the knowledge of a normalizing segment sequence, which can be composed of a single segment of any one chromosome, or of two or more segments of any two or more of chromosomes 1- 22, X, and Y.
CNV and Prenatal Diagnoses Cell-free fetal DNA and RNA circulating in maternal blood can be used for the early non-invasive prenatal sis (NIPD) of an increasing number of genetic conditions, both for pregnancy management and to aid reproductive decisionmaking.
The ce of cell-free DNA circulating in the bloodstream has been known for over 50 years. More recently, presence of small amounts of circulating fetal DNA was discovered in the maternal bloodstream during pregnancy (Lo et al., Lancet 350:485-487 [1997]). Thought to originate from dying placental cells, ee fetal DNA (cIDNA) has been shown to consists of short fragments typically fewer than 200 bp in length Chan et al., Clin Chem 50:88-92 [2004]), which can be discerned as early as 4 weeks ion (Illanes et al., Early Human Dev -566 ), and known to be cleared from the maternal circulation within hours of delivery (Lo et al., Am J Hum Genet 64:218-224 [1999]). In addition to cIDNA, fragments of cell-free fetal RNA (cfRNA) can also be discerned in the maternal bloodstream, originating from genes that are transcribed in the fetus or placenta. The extraction and subsequent analysis of these fetal genetic ts from a al blood sample offers novel opportunities for NIPD.
] The present method is a polymorphism-independent method that for use in NIPD and that does not require that the fetal cIDNA be distinguished from the maternal cIDNA to enable the ination of a fetal aneuploidy. In some embodiments, the aneuploidy is a complete chromosomal trisomy or monosomy, or a partial trisomy or monosomy. Partial oidies are caused by loss or gain of part of a chromosome, and encompass chromosomal nces resulting from unbalanced translocations, unbalanced inversions, deletions and insertions. By far, the most common known aneuploidy compatible with life is trisomy 21, i.e., Down Syndrome (DS), which is caused by the presence of part or all of chromosome 21.
Rarely, DS can be caused by an inherited or sporadic defect whereby an extra copy of all or part of chromosome 21 becomes attached to another chromosome (usually chromosome 14) to form a single aberrant chromosome. DS is associated with ectual impairment, severe learning difficulties and excess mortality caused by long-term health problems such as heart disease. Other aneuploidies with known clinical icance include Edward syndrome (trisomy 18) and Patau me (trisomy 13), which are frequently fatal within the first few months of life.
Abnormalities associated with the number of sex chromosomes are also known and include monosomy X, e.g., Turner syndrome (XO), and triple X syndrome (XXX) in female births and Kleinefelter syndrome (XXY) and XYY syndrome in male births, which are all ated with various phenotypes including sterility and reduction in intellectual . Monosomy X [45, X] is a common cause of early pregnancy loss accounting for about 7% of spontaneous abortions. Based on the liveborn ncy of 45,X (also called Turner syndrome) of 1-2/10,000, it is estimated that less than 1% of 45,X tions will survive to term. About 30% of Turners syndrome patients are mosaic with both a 45,X cell line and either a 46,XX cell line or one containing a nged X chromosome (Hook and Warburton 1983). The phenotype in a liveborn infant is relatively mild considering the high embryonic lethality and it has been hypothesized that possibly all liveborn females with Turner syndrome carry a cell line containing two sex chromosomes. Monosomy X can occur in females as 45,X or as 45,X/46XX, and in males as 45,X/46XY. Autosomal monosomies in human are generally suggested to be atible with life; however, there is quite a number of cytogenetic reports describing full my of one chromosome 21 in live born children (Vosranova let al., lar n. I: 13 [2008]; Joosten et al., Prenatal Diagn. 17:271-5 [1997]. The method described herein can be used to diagnose these and other chromosomal abnormalities prenatally.
[00414] According to some embodiments the methods disclosed herein can determine the presence or absence of chromosomal trisomies of any one of chromosomes 1-22, X and Y. Examples of chromosomal trisomies that can be detected according to the present method include without limitation y 21 (T2 l; Down Syndrome), trisomy 18 (T18; Edward's Syndrome), trisomy 16 (T16), trisomy 20 (T20), trisomy 22 (T22; Cat Eye Syndrome), trisomy 15 (TIS; Prader Willi Syndrome), y 13 (T13; Patau Syndrome), trisomy 8 (T8; Warkany Syndrome), trisomy 9, and the XXY (Kleinefelter Syndrome), XYY, or XXX trisomies.
Complete trisomies of other autosomes existing in a non-mosaic state are lethal, but can be compatible with life when present in a mosaic state. It will be appreciated that various complete trisomies, whether existing in a mosaic or non-mosaic state, and partial trisomies can be determined in fetal cIDNA according to the teachings ed herein.
[00415] Non-limiting examples of partial ies that can be determined by the present method include, but are not limited to, partial trisomy lq32-44, trisomy 9 p, trisomy 4 mosaicism, trisomy l 7p, partial trisomy 4q26-qter, partial 2p trisomy, partial trisomy lq, and/or partial trisomy 6p/monosomy 6q.
The methods disclosed herein can be also used to determine somal monosomy X, chromosomal monosomy 21, and partial monosomies such as, monosomy 13, monosomy 15, monosomy 16, monosomy 21, and monosomy 22, which are known to be involved in pregnancy miscarriage. l monosomy of chromosomes typically involved in complete oidy can also be determined by the method described herein. Non-limiting examples of deletion mes that can be determined according to the present method include syndromes caused by partial deletions of somes. Examples of partial deletions that can be determined ing to the methods described herein include without limitation partial deletions of chromosomes 1, 4, 5, 7, 11, 18, 15, 13, 17, 22 and 10, which are described in the following.
[00417] lq21.l deletion syndrome or lq21.l (recurrent) microdeletion is a rare aberration of chromosome 1. Next to the on syndrome, there is also a lq21.l duplication syndrome. While there is a part of the DNA missing with the deletion syndrome on a particular spot, there are two or three copies of a similar part of the DNA on the same spot with the duplication me. Literature refers to both the deletion and the duplication as the lq21. l copy-number variations (CNV). The lq21. l deletion can be associated with the TAR Syndrome bocytopenia with Absent radius).
Wolf-Hirschhorn syndrome (WHS) (OMIN #194190) is a uous gene deletion me associated with a hemizygous deletion of chromosome 4pl6.3. Wolf-Hirschhorn syndrome is a congenital malformation syndrome characterized by pre- and postnatal growth deficiency, developmental disability of le degree, teristic craniofacial features ('Greek warrior helmet' appearance ofthe nose, high forehead, ent glabella, hypertelorism, high-arched ws, protruding eyes, epicanthal folds, short philtrum, distinct mouth with rned corners, and micrognathia), and a seizure disorder.
Partial deletion of chromosome S, also known as Sp- or Sp minus, and S named Cris du Chat syndrome (OMIN#l234SO), is caused by a deletion of the short arm (p arm) of chromosome S (SplS.3-plS.2). Infants with this condition often have a high-pitched cry that sounds like that of a cat. The disorder is characterized by intellectual disability and delayed development, small head size (microcephaly), low birth weight, and weak muscle tone (hypotonia) in infancy, ctive facial es and possibly heart defects.
] Williams-Beuren Syndrome also known as chromosome 7ql 1.23 deletion syndrome (OMIN 1940SO) is a contiguous gene deletion syndrome resulting in a multisystem disorder caused by hemizygous deletion of l.S to 1.8 Mb on chromosome 7ql 1.23, which contains approximately 28 genes. lS [00421] Jacobsen Syndrome, also known as l lq deletion disorder, 1s a rare congenital disorder resulting from deletion of a terminal region of chromosome 11 that includes band llq24. l. It can cause intellectual disabilities, a distinctive facial appearance, and a variety of physical problems including heart defects and a bleeding disorder.
[00422] l monosomy of chromosome 18, known as monosomy l 8p is a rare chromosomal disorder in which all or part of the short arm (p) of chromosome 18 is deleted (monosomic). The disorder is lly characterized by short stature, variable degrees of mental ation, speech delays, malformations of the skull and facial (craniofacial) region, and/or additional physical abnormalities. Associated 2S craniofacial defects may vary greatly in range and ty from case to case.
] Conditions caused by s in the structure or number of copies of chromosome lS include Angelman Syndrome and Prader-Willi me, which involve a loss of gene activity in the same part of chromosome lS, the 1Sqll-ql3 region. It will be iated that several translocations and microdeletions can be asymptomatic in the carrier parent, yet can cause a major genetic disease in the offspring. For example, a healthy mother who carries the l Sq 11-q13 microdeletion can give birth to a child with Angelman syndrome, a severe neurodegenerative disorder. Thus, the methods, apparatus and systems described herein can be used to identify such a partial deletion and other deletions in the fetus.
Partial monosomy 13q is a rare chromosomal er that results when a piece of the long arm (q) of chromosome 13 is missing (monosomic). Infants born with partial monosomy 13q may exhibit low birth weight, malformations of the head and face (craniofacial region), al abnormalities (especially ofthe hands and feet), and other physical abnormalities. Mental retardation is teristic of this condition. The ity rate during infancy is high among individuals born with this disorder. Almost all cases of partial monosomy 13q occur randomly for no apparent reason (sporadic).
Magenis syndrome (SMS - OMIM #182290) is caused by a deletion, or loss of genetic material, on one copy of chromosome 17. This wellknown syndrome is associated with pmental delay, mental retardation, congenital ies such as heart and kidney defects, and neurobehavioral abnormalities such as severe sleep disturbances and self-injurious behavior. Smith- Magenis me (SMS) is caused in most cases (90%) by a 3.7-Mb interstitial deletion in chromosome l 7p 11.2. 22q11.2 deletion syndrome, also known as DiGeorge syndrome, is a syndrome caused by the deletion of a small piece of chromosome 22. The deletion (22 q11.2) occurs near the middle of the chromosome on the long arm of one of the pair of chromosome. The features of this syndrome vary widely, even among members of the same family, and affect many parts of the body. teristic signs and symptoms may include birth s such as congenital heart e, s in the palate, most commonly related to neuromuscular problems with closure (velo- pharyngeal insufficiency), learning disabilities, mild differences in facial features, and recurrent infections. Microdeletions in chromosomal region 22q11.2 are associated with a 20 to 30-fold increased risk of phrenia.
Deletions on the short arm of chromosome 10 are associated with a DiGeorge Syndrome like phenotype. Partial monosomy of chromosome lOp is rare but has been observed in a portion of patients showing features of the DiGeorge Syndrome.
In one embodiment, the methods, apparatus, and s described herein is used to determine partial monosomies including but not d to partial monosomy of chromosomes 1, 4, 5, 7, 11, 18, 15, 13, 17, 22 and 10, e.g., partial monosomy lq21. ll, partial monosomy 4pl6.3, partial monosomy 5pl5.3-pl5.2, partial monosomy 7q11.23, partial monosomy 11 q24 .1, l monosomy l 8p, partial monosomy of chromosome 15 (15ql l-ql3), partial monosomy 13q, partial monosomy l 7p 11.2, partial monosomy of chromosome 22 (22q11.2), and partial my 1Op can also be determined using the method.
Other partial monosomies that can be determined according to the methods bed herein include unbalanced translocation t(8; ll)(p23.2;pl5.5); 11 q23 microdeletion; l 7p 11.2 deletion; 22q13 .3 on; Xp22.3 microdeletion; 10pl4 deletion; 20p microdeletion, [del(22)(ql l.2ql 1.23)], 7ql 1.23 and 7q36 deletions; 1p36 deletion; 2p microdeletion; neurofibromatosis type 1 (17q11.2 microdeletion), Yq deletion ; 4pl6.3 microdeletion; lp36.2 microdeletion; l lql4 deletion; 19ql3.2 microdeletion; Rubinstein-Taybi (16 pl3.3 microdeletion); 7p21 microdeletion; Miller-Dieker syndrome (17pl3.3); and 2q37 microdeletion. Partial deletions can be small ons of part of a chromosome, or they can be microdeletions of a chromosome where the deletion of a single gene can occur.
] Several duplication syndromes caused by the duplication of part of chromosome arms have been identified (see OMIN [Online Mendelian Inheritance in Man viewed online at ncbi.nlm.nih.gov/omim]). In one embodiment, the t method can be used to determine the presence or e of duplications and/or lications of segments of any one of chromosomes 1-22, X and Y. Non-limiting examples of duplications syndromes that can be determined according to the present method include duplications of part of chromosomes 8, 15, 12, and 17, which are bed in the following. 8p23 .1 duplication me is a rare genetic er caused by a duplication of a region from human chromosome 8. This duplication syndrome has an estimated prevalence of 1 in 64,000 births and is the reciprocal of the 8p23. l deletion syndrome. The 8p23 .1 duplication is associated with a variable phenotype including one or more of speech delay, developmental delay, mild dysmorphism, with prominent forehead and arched eyebrows, and congenital heart disease (CHD).
Chromosome l 5q Duplication Syndrome (Dup l 5q) is a clinically identifiable syndrome which results from duplications of chromosome l 5q11-13 .1 Babies with Dup l 5q usually have hypotonia (poor muscle tone), growth retardation; they may be born with a cleft lip and/or palate or malformations of the heart, s or other organs; they show some degree of cognitive delay/disability (mental retardation), speech and ge delays, and sensory processing disorders. ter Killian syndrome is a result of extra #12 chromosome material. There is usually a mixture of cells (mosaicism), some with extra #12 material, and some that are normal (46 somes without the extra #12 al).
Babies with this syndrome have many problems including severe mental retardation, poor muscle tone, "coarse" facial features, and a prominent forehead. They tend to have a very thin upper lip with a thicker lower lip and a short nose. Other health problems e seizures, poor feeding, stiff joints, cataracts in adulthood, hearing loss, and heart defects. Persons with Pallister Killian have a shortened an.
[00434] duals with the genetic condition designated as dup(l7)(pl l.2pl 1.2) or dup l 7p carry extra genetic information (known as a duplication) on the short arm of chromosome 17. Duplication of chromosome l 7pl 1.2 underlies Potocki-Lupski syndrome (PTLS), which is a newly ized genetic condition with only a few dozen cases reported in the medical literature.
Patients who have this duplication often have low muscle tone, poor feeding, and failure to thrive during y, and also present with delayed development of motor and verbal milestones. Many individuals who have PTLS have difficulty with articulation and language processing. In on, patients may have behavioral characteristics similar to those seen in persons with autism or autism-spectrum disorders. Individuals with PTLS may have heart defects and sleep apnea. . A duplication of a large region in chromosome l 7pl2 that includes the gene PMP22 is known to cause Charcot-Marie Tooth disease.
CNV have been ated with stillbirths. However, due to nt limitations of conventional cytogenetics, the contribution of CNV to stillbirth is thought to be underrepresented (Harris et al., Prenatal Diagn 31:932-944 [2011]). As is shown in the examples and described elsewhere herein, the present method is capable of determining the ce of partial aneuploidies, e.g., deletions and lications of chromosome segments, and can be used to identify and determine the presence or absence of CNV that are associated with stillbirths.
Determination of CNV of al disorders In addition to the early determination of birth defects, the methods described herein can be d to the determination of any abnormality in the representation of genetic sequences within the genome. A number of abnormalities in the representation of genetic sequences within the genome have been associated with various pathologies. Such ogies include, but are not limited to cancer, infectious and autoimmune diseases, diseases ofthe nervous system, metabolic and/or cardiovascular diseases, and the like.
Accordingly in various embodiments use of the methods described herein in the diagnosis, and/or ring, and or treating such pathologies is contemplated. For e, the methods can be applied to determining the presence or absence of a disease, to monitoring the progression of a disease and/or the efficacy of a treatment regimen, to determining the ce or absence of nucleic acids of a pathogen e.g. virus; to ining chromosomal abnormalities associated with graft versus host disease (GVHD), and to determining the bution of individuals in forensic analyses.
CNVs in Cancer
[00438] It has been shown that blood plasma and serum DNA from cancer patients contains measurable ties oftumor DNA, that can be recovered and used as surrogate source of tumor DNA, and tumors are characterized by aneuploidy, or inappropriate numbers of gene ces or even entire chromosomes. The determination of a difference in the amount of a given sequence i.e. a sequence of interest, in a sample from an individual can thus be used in the prognosis or diagnosis of a medical condition. In some embodiments, the present method can be used to determine the presence or absence of a chromosomal aneuploidy in a patient suspected or known to be suffering from cancer.
Some entations herein e methods for detecting cancer, tracking therapeutic se and minimal residual disease based on circulating cIDNA samples using shallow sequencing of the samples with paired-end methodology and using fragment size information available from paired-end reads to identify ce of differentially-methylated tic DNA from cancer cells in the background of normal cells. It has been shown that tumor-derived cIDNA are shorter than non-tumor-derived cIDNA in some cancers. Therefore the size-based method described herein can be used to determine CNV including aneuploidies associated with these cancers, enabling (a) ion of tumor present in a screening or diagnostic setting; (b) monitoring response to therapy; (c) monitoring minimal residual e.
In certain ments the oidy is characteristic of the genome of the subject and results in a lly increased predisposition to a cancer. In certain ments the aneuploidy is characteristic of particular cells (e.g., tumor cells, proto-tumor neoplastic cells, etc.) that are or have an increased predisposition to neoplasia. Particular aneuploidies are associated with particular cancers or predispositions to particular cancers as described below. In some embodiments, a very shallow paired-end sequencing approach can be used to detect I monitor cancer presence in a cost-effective way.
Accordingly, various embodiments of the methods described herein provide a determination of copy number variation of sequence(s) of interest e.g. clinically-relevant sequence(s), in a test sample from a subject where certain variations in copy number e an indicator ofthe presence and/or a predisposition to a cancer. In certain embodiments the sample comprises a mixture of nucleic acids is derived from two or more types of cells. In one embodiment, the mixture ofnucleic acids is derived from normal and cancerous cells derived from a subject suffering from a medical condition e.g. cancer.
[00442] The development of cancer is often accompanied by an alteration in number of whole somes i.e. complete chromosomal aneuploidy, and/or an alteration in the number of segments of chromosomes i.e. partial aneuploidy, caused by a process known as chromosome instability (CIN) (Thoma et al., Swiss Med Weekly 2011: 141:w13170). It is believed that many solid tumors, such as breast cancer, ss from initiation to metastasis through the lation of several genetic tions. [Sato et al., Cancer Res., 50: 7184-7189 [1990]; Jongsma et al., J Clin : Mol Path 55:305-309 [2002])]. Such genetic aberrations, as they accumulate, may confer erative advantages, genetic instability and the attendant ability to evolve drug resistance rapidly, and enhanced angiogenesis, proteolysis and asis. The c aberrations may affect either recessive "tumor suppressor genes" or ntly acting oncogenes. Deletions and recombination leading to loss of heterozygosity (LOH) are believed to play a major role in tumor progression by uncovering mutated tumor suppressor alleles. cIDNA has been found in the circulation of patients sed with malignancies including but not limited to lung cancer (Pathak et al. Clin Chem 52: 1833-1842 [2006]), prostate cancer (Schwartzenbach et al. Clin Cancer Res : 1032-8 [2009]), and breast cancer (Schwartzenbach et al. available online at -cancer-research.com/content/1 l/5/R71 [2009]). Identification of genomic instabilities ated with cancers that can be determined in the circulating cIDNA in cancer patients is a potential diagnostic and prognostic tool. In one embodiment, methods described herein are used to ine CNV of one or more sequence(s) of interest in a sample, e.g., a sample comprising a mixture of nucleic acids d from a subject that is suspected or is known to have cancer e.g. carcinoma, sarcoma, lymphoma, leukemia, germ cell tumors and blastoma. In one embodiment, the sample is a plasma sample derived ssed) from peripheral blood that may comprise a mixture of cIDNA derived from normal and cancerous cells. In another embodiment, the biological sample that is needed to determine whether a CNV is present is derived from a cells that, if a cancer is present, comprise a mixture of cancerous and noncancerous cells from other biological tissues including, but not limited to biological fluids such as serum, sweat, tears, , urine, sputum, ear flow, lymph, saliva, cerebrospinal fluid, ravages, bone marrow suspension, vaginal flow, transcervical lavage, brain fluid, ascites, milk, secretions of the respiratory, inal and genitourinary tracts, and leukophoresis samples, or in tissue biopsies, swabs, or smears. In other embodiments, the biological sample is a stool (fecal) sample.
The methods described herein are not limited to the analysis of cIDNA.
It will be recognized that similar es can be performed on cellular DNA samples.
In various embodiments the sequence(s) of interest comprise nucleic acid sequence(s) known or is suspected to play a role in the development and/or progression of the cancer. Examples of a sequence of interest include c acids sequences e.g. complete chromosomes and/or segments of somes, that are amplified or deleted in cancerous cells as described below.
Total CNVnumber and risk for cancer.
Common cancer SNPs - and by analogy common cancer CNVs may each confer only a minor increase in disease risk. However, collectively they may cause a substantially elevated risk for s. In this regard it is noted that germline gains and losses of large DNA segments have been reported as factors predisposing duals to lastoma, prostate and colorectal cancer, breast cancer, and BRCAI-associated ovarian cancer (see, e.g., Krepischi et al. Breast Cancer Res., 14: R24 ; Diskin et al. Nature 2009, 459:987-991; Liu et al. Cancer Res 2009, 69: 2176-2179; Lucito et al. Cancer Biol Ther 2007, 6: 1592-1599; Thean et al. Genes Chromosomes Cancer 2010, 49:99-106; Venkatachalam et al. Int J Cancer 2011, 129: 1635-1642; and Yoshihara et al. Genes somes Cancer 2011, 50: 167-177).
It is noted that CNVs frequently found in the healthy population (common CNVs) are believed to have a role in cancer etiology (see, e.g., Shlien and Malkin (2009) Genome Medicine, 1(6): 62). In one study testing the hypothesis that common CNVs are associated with malignancy (Shlien et al. Proc Natl Acad Sci USA 2008, 105:11264-11269) a map of every known CNV whose locus coincides with that of bona fide cancer-related genes (as catalogued by Higgins et al. Nucleic Acids Res 2007, 35:D721-726) was created. These were termed "cancer CNVs". In an initial analysis (Shlien et al. Proc Natl Acad Sci USA 2008, 105: 11264-11269), 770 healthy genomes were evaluated using the Affymetrix SOOK array set, which has an average inter-probe distance of 5.8 kb. As CNVs are generally thought to be depleted in gene regions (Redon et al. (2006) Nature 2006, 444:444-454), it was surprising to find 49 cancer genes that were ly encompassed or overlapped by a CNV in more than one person in a large reference population. In the top ten genes, cancer CNVs could be found in four or more people.
It is thus believed that CNV frequency can be used as a measure ofrisk for cancer (see, e.g., U.S. Patent Publication No: 2010/0261183 Al). The CNV frequency can be determined simply by the constitutive genome of the organism or it can ent a on derived from one or more tumors (neoplastic cells) if such are present.
In certain embodiments a number of CNVs in a test sample (e.g., a sample comprising a constitutional (germline) nucleic acid) or a mixture of nucleic acids (e.g., a ne nucleic acid and nucleic ) d from neoplastic cells) is determined usmg the methods described herein for copy number variations.
Identification of an increased number of CNVs in the test sample, e.g., in comparison to a reference value is indicative of a risk of or pre-disposition for cancer in the t. It will be appreciated that the reference value may vary with a given population. It will also be iated that the absolute value of the se in CNV frequency will vary depending on the resolution of the method utilized to determine CNV frequency and other parameters. Typically, an increase in CNV frequency of at least about 1.2 times the reference value been determined to tive of risk for cancer (see, e.g., U.S. Patent ation No: 2010/0261183 Al), for example an increase in CNV frequency of at least or about 1.5 times the reference value or greater, such as 2-4 times the reference value is an indicator of an increased risk of cancer (e.g., as compared to the normal healthy reference population).
A determination of structural ion in the genome of a mammal in comparison to a reference value is also believed to be indicative of risk of cancer. In this context, in one embodiment, the term tural variation" is can be defined as the CNV frequency in a mammal multiplied by the average CNV size (in bp) in the mammal. Thus, high structural variation scores will result due to increased CNV frequency and/or due to the occurrence of large genomic nucleic acid deletions or duplications. Accordingly, in certain embodiments a number of CNVs in a test sample (e.g., a sample comprising a constitutional (germline) nucleic acid) is ined using the methods described herein to determine size and number of copy number variations. In certain embodiments a total structural variation score within genomic DNA of greater than about I megabase, or greater than about 1.1 megabases, or greater than about 1.2 megabases, or greater than about 1.3 megabases, or greater than about 1.4 megabases, or r than about 1.5 megabases, or greater than about 1.8 megabases, or greater than about 2 megabases of DNA is indicative of risk of cancer.
It is believed these s e a e of the risk of any cancer including but not limited to, acute and chronic leukemias, mas, numerous solid tumors of mesenchymal or epithelial tissue, brain, breast, liver, stomach, colon cancer, B cell lymphoma, lung cancer, a bronchus cancer, a colorectal cancer, a prostate cancer, a breast cancer, a pancreas cancer, a stomach cancer, an ovarian , a urinary bladder cancer, a brain or central nervous system cancer, a peripheral nervous system cancer, an esophageal , a cervical cancer, a melanoma, a uterine or trial cancer, a cancer of the oral cavity or x, a liver cancer, a kidney cancer, a biliary tract cancer, a small bowel or ix cancer, a salivary gland cancer, a thyroid gland , a adrenal gland cancer, an arcoma, a chondrosarcoma, a liposarcoma, a testes , and a ant fibrous histiocytoma, and other cancers.
Full chromosome aneuploidies.
As ted above, there exists a high frequency of aneuploidy m cancer. In certain studies examining the prevalence of somatic copy number alterations (SCNAs) in cancer, it has been discovered that one-quarter of the genome of a typical cancer cell is affected either by whole-arm SCNAs or by the wholechromosome SCNAs of aneuploidy (see, e.g., Beroukhim et al. Nature 463: 899-905 ). Whole-chromosome alterations are recurrently observed in several cancer types. For example, the gain of chromosome 8 is seen in 10-20% of cases of acute myeloid leukaemia (AML), as well as some solid tumours, including Ewing's Sarcoma and desmoid tumours (see, e.g., Barnard et al. Leukemia IO: 5-12 [1996]; Maurici et al. Cancer Genet. Cytogenet. 100: 106-110 [1998]; Qi et al. Cancer Genet. Cytogenet. 92: 147-149 [1996]; Barnard, D. R. et al. Blood 100: 427-434 ; and the like. Illustrative, but non-limiting list of some gains and losses in human cancers are shown in Table 2.
TABLE 2. Illustrative specific, recurrent chromosome gains and losses in human cancer (see, e.g., Gordon et al. (2012) Nature Rev.
Genetics, 13: 189-203).
Chromosome Gains Losses Cancer Type Cancer Type I Multiple myeloma Adenocarcinoma (kidney) Adenocarcinoma (breast) 2 Hepatoblastoma ssarcoma 3 Multiple myeloma Melanoma Diffuse large B-cell lymphoma Adenocarcinoma (kidney) 4 Acute lymphoblastic leukaemia Adenocarcinoma (kidney) le myeloma Adenocarcinoma y) 6 Acute lymphoblastic leukaemia Adenocarcinoma (kidney) Wilms'tumour 7 Adenocarcinoma (kidney) Acute myeloid leukaemia Adenocarcinoma (intestine) Juvenile myelomonocytic leukaemia 8 Acute myeloid leukaemia Adenocarcinoma (kidney) Chronic d leukaemia Ewing'ssarcoma 9 Multiple myeloma Polycythaemia vera Acute lymphoblastic leukaemia Astrocytoma arcinoma (uterus) Multiple a 11 Multiple myeloma 12 Chronic lymphocytic leukaemia Multiple myeloma Wilms'tumor 13 Acute myeloid leukaemia Multiple myeloma Wilms'tumor 14 Acute lymphoblastic leukaemia Adenocarcinoma (kidney) Meningioma Multiple myeloma 16 Adenocarcinoma (kidney) Multiple myeloma 17 Adenocarcinoma (kidney) Acute lymphoblastic leukaemia 18 Acute lymphoblastic leukaemia Adenocarcinoma (kidney) Wilms'tumour 19 Multiple myeloma Adenocarcinoma (Breast) Chronic d leukaemia Meningioma Hepatoblastoma Adenocarcinoma (kidney) 21 Acute lymphoblastic mia Acute megakaryoblastic leukaemia 22 Acute lymphoblastic leukaemia Meningioma x Acute lymphoblastic mia Follicular lymphoma In various embodiments, the methods described herein can be used to detect and/or quantify whole some aneuploidies that are associated with cancer generally, and/or that are associated with particular cancers. Thus, for e, in certain embodiments, detection and/or fication of whole chromosome aneuploidies characterized by the gains or losses shown in Table 2 are contemplated.
Arm level somal segment copv number variations.
Multiple studies have reported patterns of arm-level copy number ions across large numbers of cancer specimens (Lin et al. Cancer Res 68, 664- 673 (2008); George et al. PLoS ONE 2, e255 (2007); Demichelis et al. Genes Chromosomes Cancer 48: 366-380 (2009); Beroukhim et al. Nature. 463(7283): 899- 905 [2010]). It has additionally been observed that the frequency of arm-level copy number variations ses with the length of chromosome arms. Adjusted for this trend, the majority of chromosome arms t strong ce of preferential gain or loss, but rarely both, across multiple cancer lineages (see, e.g., Beroukhim et al.
Nature. 463(7283): 899-905 [2010]).
Accordingly, in one embodiment, methods described herein are used to determine arm level CNVs (CNVs comprising one chromosomal arm or substantially one chromosomal arm) in a sample. The CNVs can be determined in a CNVs in a test sample comprising a constitutional (germline) nucleic acid and the arm level CNVs can be identified in those constitutional nucleic acids. In certain embodiments arm level CNVs are identified (if present) in a sample comprising a mixture of nucleic acids (e.g., nucleic acids derived from normal and nucleic acids derived from neoplastic cells). In certain embodiments the sample is derived from a subject that is suspected or is known to have cancer e.g. carcinoma, sarcoma, lymphoma, ia, germ cell tumors, blastoma, and the like. In one embodiment, the sample is a plasma sample d (processed) from peripheral blood that may comprise a mixture of cIDNA d from normal and ous cells. In another embodiment, the biological sample that is used to determine whether a CNV is present is derived from a cells that, if a cancer is present, comprise a mixture of cancerous and non-cancerous cells from other biological tissues including, but not limited to biological fluids such as serum, sweat, tears, , urine, sputum, ear flow, lymph, saliva, cerebrospinal fluid, ravages, bone marrow suspension, vaginal flow, transcervical lavage, brain fluid, ascites, milk, secretions of the respiratory, intestinal and genitourinary tracts, and horesis samples, or in tissue biopsies, swabs, or smears. In other embodiments, the biological sample is a stool ) sample.
In various embodiments the CNVs identified as indicative of the presence of a cancer or an increased risk for a cancer include, but are not limited to the arm level CNVs listed in Table 3. As illustrated in Table 3 certain CNVs that se a substantial arm-level gain are tive of the presence of a cancer or an increased risk for a certain cancers. Thus, for example, a gain in lq is indicative of the presence or sed risk for acute lymphoblastic leukemia (ALL), breast , GIST, HCC, lung NSC, medulloblastoma, melanoma, MPD, ovarian cancer, and/or prostate cancer. A gain in 3q is indicative of the presence or increased risk for WO 36059 Esophageal Squamous cancer, Lung SC, and/or MPD. A gain in 7q is indicative of the presence or increased risk for colorectal cancer, , HCC, lung NSC, medulloblastoma, melanoma, prostate cancer, and/or renal cancer. A gain in 7p is indicative of the presence or sed risk for breast cancer, colorectal cancer, esophageal adenocarcinoma, glioma, HCC, Lung NSC, medulloblastoma, melanoma, and/or renal cancer. A gain in 20q is indicative of the presence or increased risk for breast cancer, colorectal cancer, dedifferentiated liposarcoma, esophageal adenocarcinoma, esophageal squamous, glioma cancer, HCC, lung NSC, melanoma, n cancer, and/or renal cancer, and so forth.
[00456] Similarly as illustrated in Table 3 certain CNVs that e a substantial arm-level loss are indicative ofthe presence of and/or an increased risk for certain cancers. Thus, for example, a loss in Ip is indicative of the presence or increased risk for intestinal stromal tumor. A loss in 4q is indicative of the presence or increased risk for colorectal cancer, esophageal adenocarcinoma, lung sc, melanoma, ovarian cancer, and/or renal cancer. a loss in l 7p is indicative of the ce or increased risk for breast cancer, colorectal cancer, esophageal adenocarcinoma, HCC, lung NSC, lung SC, and/or ovarian cancer, and the like.
TABLE 3. Significant vel chromosomal segment coov number tions in each of 16 cancer subtypes (breast, ctal, dedifferentiated liposarcoma, esophageal adenocarcinoma, esophageal squamous, GIST (gastrointestinal stromal tumor), glioma, HCC (hepatocellular oma), lung NSC, lung SC, medulloblastoma, melanoma, MPD (myeloproliferative disease), ovarian, prostate, acute lymphoblastic leukemia (ALL), and renal) (see, e.g., Beroukhim et al. Nature (2010) 83): 899-905).
Arm Cancer Types Cancer Types Known Significantly Gained In Significantly Lost In Oncogene/Tumor Suooressor Gene Ip ---- GIST lq ALL, Breast, GIST, HCC, ---- LungNSC, Medulloblastoma, Melanoma, MPD, Ovarian, Prostate 3p ---- Esophageal Squamous, VHL Lung NSC, Lung SC, Renal 3q Esophageal Squamous, ---- Lung SC, MPD 4p ALL Breast, Esophageal Adenocarcinoma, Renal 4q ALL Colorectal, Esophageal Adenocarcinoma, Lung SC, Melanoma, Ovarian, Renal Sp Esophageal us, ---- TERT HCC, Lung NSC, Lung SC, Renal Sq HCC, Renal Esophageal AFC Adenocarcinoma, Lung 6p ALL, HCC, Lung NSC, ---- Melanoma 6q ALL Melanoma, Renal 7p Breast, Colorectal, ---- EGFR Esophageal Adenocarcinoma, Glioma, HCC, Lung NSC, Medulloblastoma, Melanoma, Renal 7q Colorectal, Glioma, HCC, ---- BRAF,MET LungNSC, Medulloblastoma, Melanoma, Prostate, Renal 8p ALL,MPD , HCC, Lung NSC, Medulloblastoma, Prostate, Renal 8q ALL, Breast, Colorectal, Medulloblastoma MYC Esophageal Adenocarcinoma, Esophageal Squamous, HCC, LungNSC, MPD, Ovarian, te 9p MPD ALL, , Esophageal CDKN2AIB Adenocarcinoma, Lung NSC, Melanoma, Ovarian, Renal 9q ALL,MPD Lung NSC, Melanoma, n, Renal IOp ALL Glioma, Lung SC, lOq ALL Glioma, Lung SC, PTEN Medulloblastoma, Melanoma llp ---- Medulloblastoma WTI llq ---- Dedifferentiated ATM Liposarcoma, Medulloblastoma, Melanoma 12p Colorectal, Renal ---- KRAS 12q Renal ---- 13q Colorectal Breast, Dedifferentiated RBllBRCA2 Liposarcoma, Glioma, Lung NSC, Ovarian 14q ALL, Lung NSC, Lung SC, GIST, Melanoma, Renal Prostate 15q ---- GIST, Lung NSC, Lung SC, n 16p Breast ---- 16q ---- Breast, HCC, oblastoma, Ovarian, Prostate 17p ALL Breast, Colorectal, TP53 Esophageal Adenocarcinoma, HCC, Lung NSC, Lung SC, Ovarian 17q ALL, HCC, Lung NSC, Breast, Ovarian ERBB2, oblastoma NFllBRCAI 18p ALL, Medulloblastoma Colorectal, Lung NSC 18q ALL, Medulloblastoma Colorectal, Esophageal SMAD2, SMAD4 Adenocarcinoma, Lung 19p Glioma Esophageal Adenocarcinoma, Lung NSC, Melanoma, Ovarian 19q Glioma, Lung SC Esophageal Adenocarcinoma, Lung 20p Breast, Colorectal, ---- geal Adenocarcinoma, Esophageal Squamous, GIST, , HCC, Lung NSC, Melanoma, Renal 20q Breast, Colorectal, ---- Dedifferentiated Liposarcoma, Esophageal Adenocarcinoma, Esophageal Squamous, Glioma, HCC, Lung NSC, Melanoma, Ovarian, Renal 2lq ALL, GIST, MPD ---- 22q Melanoma , Colorectal, NF2 Dedifferentiated Liposarcoma, Esophageal Adenocarcinoma, GIST, 2016/067886 Lung NSC, Lung SC, I Ovarian, Prostate The examples of associations between arm level copy number variations are intended to be illustrative and not limiting. Other arm level copy number variations and their cancer associations are known to those of skill in the art.
Smaller. e.g.. focal. copy number variations.
[00458] As indicated above, in certain embodiments, the s described herein can be used to determine the presence or absence of a chromosomal amplification. In some embodiments, the chromosomal amplification is the gain of one or more entire chromosomes. In other embodiments, the chromosomal ication is the gain of one or more segments of a chromosome. In yet other embodiments, the chromosomal amplification is the gain of two or more segments of two or more chromosomes. In s embodiments, the chromosomal amplification can involve the gain of one or more oncogenes.
Dominantly acting genes associated with human solid tumors typically exert their effect by overexpression or altered expression. Gene ication is a common mechanism leading to upregulation of gene expression. Evidence from cytogenetic studies indicates that significant amplification occurs in over 50% of human breast cancers. Most notably, the amplification of the proto-oncogene human epidermal growth factor receptor 2 (HER2) located on chromosome 17 (17(17q21- q22)), results in overexpression of HER2 ors on the cell surface leading to excessive and dysregulated signaling in breast cancer and other malignancies (Park et al., Clinical Breast Cancer 8:392-401 [2008]). A variety of nes have been found to be amplified in other human malignancies. Examples ofthe amplification of cellular oncogenes in human tumors include amplifications of: c-myc m promyelocytic leukemia cell line HL60, and in small-cell lung carcinoma cell lines, N-myc in primary neuroblastomas (stages III and IV), neuroblastoma cell lines, retinoblastoma cell line and primary tumors, and cell lung carcinoma lines and tumors, L-myc in small-cell lung carcinoma cell lines and tumors, c-myb in acute d leukemia and in colon carcinoma cell lines, c-erbb in epidermoid carcinoma cell, and primary gliomas, c-K-ras-2 in primary omas of lung, colon, bladder, and rectum, N-ras in mammary carcinoma cell line (Varmus H., Ann Rev Genetics I8: 553-6I2 (I984) [cited in Watson et al., Molecular Biology of the Gene (4th ed.; in/Cummings Publishing Co. I987)].
Duplications of oncogenes are a common cause of many types of cancer, as is the case with P70-S6 Kinase I ication and breast cancer. In such cases the genetic duplication occurs in a c cell and s only the genome of the cancer cells themselves, not the entire organism, much less any subsequent offspring. Other examples of oncogenes that are amplified in human cancers include MYC, ERBB2 (EFGR), CCNDI (Cyclin DI), FGFRI and FGFR2 in breast cancer, MYC and ERBB2 in cervical cancer, HRAS, KRAS, and MYB in ctal cancer, I0 MYC, CCND I and MDM2 in esophageal , CCNE, KRAS and MET in gastric cancer, ERBBI, and CDK4 in glioblastoma, CCNDI, ERBBI, and MYC in head and neck cancer, CCND I in hepatocellular cancer, MYCB in neuroblastoma, MYC, ERBB2 and AKT2 in ovarian cancer, MDM2 and CDK4 in sarcoma, and MYC in small cell lung cancer. In one embodiment, the present method can be used to I5 determine the presence or absence of amplification of an oncogene associated with a cancer. In some embodiments, the amplified oncogene is associated with breast , cervical cancer, colorectal cancer, esophageal cancer, gastric cancer, glioblastoma, head and neck cancer, hepatocellular cancer, neuroblastoma, ovanan cancer, sarcoma, and small cell lung .
[00461] In one embodiment, the present method can be used to determine the presence or absence of a chromosomal deletion. In some embodiments, the chromosomal deletion is the loss of one or more entire chromosomes. In other embodiments, the chromosomal deletion is the loss of one or more segments of a chromosome. In yet other embodiments, the chromosomal deletion is the loss of two or more segments of two or more chromosomes. The somal deletion can involve the loss of one or more tumor suppressor genes.
Chromosomal deletions involving tumor suppressor genes are believed to play an important role in the development and progression of solid tumors. The retinoblastoma tumor ssor gene (Rb-I), located in some 13qI4, is the most extensively characterized tumor suppressor gene. The Rb-I gene product, a I05 kDa r phosphoprotein, apparently plays an important role in cell cycle regulation (Howe et al., Proc Natl Acad Sci (USA) 87:5883-5887 [I990]). d or lost expression of the Rb protein is caused by inactivation of both gene alleles either WO 36059 through a point mutation or a chromosomal deletion. Rb-i gene alterations have been found to be present not only in retinoblastomas but also in other malignancies such as osteosarcomas, small cell lung cancer (Rygaard et al., Cancer Res 50: 5312-5317 [1990)]) and breast cancer. Restriction fragment length polymorphism (RFLP) studies have indicated that such tumor types have frequently lost heterozygosity at 13q ting that one of the Rb-I gene alleles has been lost due to a gross somal deletion (Bowcock et al., Am J Hum Genet, 46: 12 [1990]).
Chromosome I abnormalities including duplications, deletions and unbalanced translocations involving some 6 and other partner chromosomes indicate that regions of chromosome 1, in particular lq21-lq32 and lpl 1-13, might harbor nes or tumor ssor genes that are pathogenetically relevant to both chronic and advanced phases of myeloproliferative neoplasms (Caramazza et al., Eur J Hematol 84: 191-200 [20 IO]). Myeloproliferative neoplasms are also associated with deletions of chromosome 5. Complete loss or interstitial deletions of chromosome 5 are the most common karyotypic abnormality in myelodysplastic syndromes (MDSs). Isolated )/5q- MDS patients have a more favorable prognosis than those with additional karyotypic defects, who tend to develop myeloproliferative neoplasms (MPNs) and acute myeloid leukemia. The frequency of unbalanced chromosome 5 deletions has led to the idea that Sq harbors one or more tumor-suppressor genes that have fundamental roles in the growth control of hematopoietic stem/progenitor cells (HSCs/HPCs). Cytogenetic mapping of commonly deleted regions (CDRs) centered on 5q3 l and 5q32 identified candidate tumor-suppressor genes, ing the ribosomal subunit RPS 14, the transcription factor Egrl/Krox20 and the cytoskeletal remodeling protein, alpha-catenin (Eisenmann et al., Oncogene 28:3429-3441 ). Cytogenetic and typing studies of fresh tumors and tumor cell lines have shown that c loss from several ct regions on chromosome 3p, including 3p25, 3p21-22, 3p21.3, 3pl2-13 and 3pl4, are the earliest and most frequent c abnormalities involved in a wide spectrum of major epithelial s of lung, , kidney, head and neck, ovary, cervix, colon, pancreas, esophagus, bladder and other organs. Several tumor suppressor genes have been mapped to the chromosome 3p region, and are thought that interstitial ons or promoter hypermethylation precede the loss of the 3p or the entire chromosome 3 in the development of carcinomas (Angeloni D., Briefings Functional Genomics 6: 19-39 [2007]).
Newborns and children with Down syndrome (DS) often present with ital transient leukemia and have an increased risk of acute myeloid ia and acute blastic ia. some 21, harboring about 300 genes, may be involved in numerous structural aberrations, e.g., translocations, deletions, and amplifications, in leukemias, mas, and solid tumors. Moreover, genes located on chromosome 21 have been identified that play an important role in tumorigenesis.
Somatic numerical as well as structural chromosome 21 aberrations are associated with leukemias, and specific genes ing RUNXl, 2, and TFF, which are located in 2lq, play a role in tumorigenesis (Fonatsch C Gene Chromosomes Cancer 49:497-508 [2010]).
In view of the foregoing, in vanous embodiments the methods described herein can be used to determine the segment CNVs that are known to comprise one or more oncogenes or tumor suppressor genes, and/or that are known to be associated with a cancer or an increased risk of cancer. In certain embodiments, the CNVs can be determined in a test sample comprising a constitutional (germline) nucleic acid and the segment can be identified in those constitutional c acids. In certain embodiments segment CNVs are identified (if present) in a sample comprising a mixture of nucleic acids (e.g., nucleic acids derived from normal and nucleic acids derived from neoplastic cells). In certain embodiments the sample is derived from a subject that is suspected or is known to have cancer e.g. carcinoma, sarcoma, lymphoma, leukemia, germ cell tumors, blastoma, and the like. In one embodiment, the sample is a plasma sample derived (processed) from peripheral blood that may comprise a mixture of cIDNA derived from normal and cancerous cells. In another embodiment, the biological sample that is used to determine whether a CNV is t is derived from a cells that, if a cancer is present, comprises a mixture of cancerous and non-cancerous cells from other biological tissues including, but not limited to biological fluids such as serum, sweat, tears, sputum, urine, sputum, ear flow, lymph, saliva, cerebrospinal fluid, ravages, bone marrow sion, vaginal flow, transcervical , brain fluid, s, milk, secretions of the atory, intestinal and genitourinary tracts, and leukophoresis samples, or in tissue biopsies, swabs, or smears. In other embodiments, the biological sample is a stool (fecal) sample.
The CNVs used to determine presence of a cancer and/or increased risk for a cancer can comprise amplification or deletions.
In various embodiments the CNVs identified as indicative of the presence of a cancer or an increased risk for a cancer include one or more of the amplifications shown in Table 4.
TABLE 4. Illustrative, but non-limiting chromosomal segments characterized by amplifications that are associated with cancers.
Cancer types listed are those identified in Beroukhim et al. Nature 18: 463: 899-905.
Peak region Length (Mb) Cancer types identified in this is but not prior publications chrl: 119996566- 0.228 Breast, Lung SC, Melanoma 120303234 chrl: 148661965- 0.35 Breast, Dedifferentiated liposarcoma, 149063439 Esophageal adenocarcinoma, cellular, Lung SC, Melanoma, Ovarian, Prostate, Renal chrl: 1-5160566 4.416 Esophageal adenocarcinoma, Ovarian chrl: 158317017- 1.627 Dedifferentiated rcoma, Esophageal 159953843 adenocarcinoma, Prostate, Renal chrl: 478- 0.889 Colorectal, erentiated liposarcoma, 170484405 Prostate, Renal chrl :201678483- 1.471 Prostate chrl :241364021- 5.678 Lung NSC, Melanoma, n 247249719 chrl :39907605- 0.319 Acute lymphoblastic leukemia, Breast, 40263248 Lung NSC, Lung SC chrl :58658784- 1.544 Breast, Dedifferentiated liposarcoma, 44 Lung SC chr3: 984- 3.496 Breast, Esophageal adenocarcinoma, 173604597 Glioma chr3: 178149984- 21.123 Esophageal squamous, Lung NSC 199501827 chr3: 86250885- 8.795 Lung SC, Melanoma 95164178 chr4: 54471680- 1.449 LungNSC 55980061 chr5: 1212750-1378766 0.115 Dedifferentiated liposarcoma chr5: 192- 6.124 Breast, Lung NSC 180857866 chr5:45312870- 4.206 Lung SC 49697231 chr6: 1-23628840 23.516 Esophageal adenocarcinoma chr6: 135561194- 0.092 Breast, Esophageal adenocarcinoma 135665525 chr6:43556800- 0.72 Esophageal adenocarcinoma, 44361368 Hepatocellular, Ovarian chr6:63255006- 1.988 Esophageal adenocarcinoma, Lung NSC 65243766 chr7:115981465- 0.69 Esophageal adenocarcinoma, Lung NSC, 116676953 Melanoma, Ovarian chr7:54899301- 0.363 Esophageal arcinoma, Esophageal 55275419 squamous chr7:89924533- 9.068 Breast, Esophageal adenocarcinoma, 98997268 Esophageal squamous, Ovarian chr8: 101163387- 2.516 Lung NSC, Melanoma, Ovarian 103693879 chr8: 116186189- 4.4 Breast, Hepatocellular, Lung NSC, 120600761 Ovarian chr8: 128774432- 0.009 Esophageal adenocarcinoma, Esophageal 128849112 squamous, Hepatocellular, Lung SC, Medulloblastoma, Myeloproliferative disorder, Ovarian chr8:140458177- 5.784 Lung NSC, oblastoma, Melanoma, 146274826 Ovarian chr8:38252951- 0.167 Colorectal, Esophageal adenocarcinoma, 38460772 Esophageal squamous chr8:42006632- 0.257 Esophageal adenocarcinoma, Lung NSC, 42404492 Lung SC, Ovarian, Prostate chr8:81242335- 0.717 Breast, Melanoma 81979194 chr9: 137859478- 2.29 Colorectal, Dedifferentiated liposarcoma 140273252 74560456- 7.455 , Ovarian, Prostate 82020637 chrll:101433436- 0.683 Lung NSC, Lung SC 102134907 chrll :32027116- 5.744 Breast, Dedifferentiated liposarcoma, 37799354 Lung NSC, Lung SC chrl 1:69098089- 0.161 Dedifferentiated liposarcoma, Esophageal 69278404 adenocarcinoma, Hepatocellular, Lung SC, Ovarian chrl 1:76699529- 1.286 erentiated liposarcoma, Esophageal 78005085 adenocarcinoma, Lung SC, Ovarian chr12: 1-1311104 1.271 LungNSC chrl2:25 l 89655- 0.112 Acute lymphoblastic leukemia, 25352305 Esophageal adenocarcinoma, geal squamous, Ovarian chr12: 23- 1.577 Acute blastic ia, 32594050 Colorectal, Esophageal adenocarcinoma, Esophageal squamous, Lung NSC, Lung chrl2:38788913- 3.779 Breast, ctal, Dedifferentiated 42596599 liposarcoma, Esophageal squamous, Lung NSC, Lung SC chrl2:56419524- 0.021 Dedifferentiated liposarcoma, Melanoma, 56488685 Renal chr12: 64461446- 0.041 Dedifferentiated liposarcoma, Renal 64607139 chrl2:66458200- 0.058 Dedifferentiated liposarcoma, Esophageal 66543552 squamous, Renal chrl2:67440273- 0.067 Breast, erentiated liposarcoma, 67566002 Esophageal squamous, Melanoma, Renal chrl2:68249634- 0.06 Breast, Dedifferentiated liposarcoma, 68327233 Esophageal squamous, Renal 70849987- 0.036 erentiated rcoma, Renal 70966467 chrl2:72596017- 0.23 Renal 73080626 chrl2:76852527- 0.158 Dedifferentiated liposarcoma 77064746 chrl2:85072329- 0.272 Dedifferentiated liposarcoma 85674601 chrl2:95089777- 0.161 Dedifferentiated liposarcoma 95350380 chr13: 108477140- 1.6 Breast, Esophageal adenocarcinoma, 110084607 Lung NSC, Lung SC chr13: 1-40829685 22.732 Acute lymphoblastic leukemia, Esophageal adenocarcinoma chr13 :89500014- 3.597 Breast, geal adenocarcinoma, 93206506 Medulloblastoma chr14: 106074644- 0.203 geal squamous 106368585 chrl4: 1-23145193 3.635 Acute lymphoblastic leukemia, Esophageal us, cellular, Lung SC chrl4:35708407- 0.383 Breast, Esophageal adenocarcinoma, 36097605 Esophageal squamous, cellular, Prostate 96891354- 0.778 Breast, Colorectal, Esophageal 97698742 adenocarcinoma, Lung NSC, Medulloblastoma, Melanoma chrl7: 23- 0.815 Breast, Hepatocellular 19933105 chrl7:22479313- 0.382 Breast, Lung NSC 22877776 chrl7:24112056- 0.114 Breast, Lung NSC 24310787 chrl7:35067383- 0.149 Colorectal, Esophageal adenocarcinoma, 35272328 Esophageal squamous chrl7:44673157- 0.351 Melanoma 45060263 chrl7:55144989- 0.31 Lung NSC, oblastoma, Melanoma, 55540417 Ovarian chrl7:62318152- 1.519 Breast, Lung NSC, Melanoma, n 63890591 chrl7:70767943- 0.537 , Lung NSC, Melanoma, Ovarian 71305641 chrl8: 17749667- 5.029 Colorectal, Esophageal adenocarcinoma, 22797232 Ovarian chrl9:3497553 l- 0.096 , Esophageal adenocarcinoma, 35098303 Esophageal squamous chrl9:43 l 77306- 2.17 Lung NSC, Ovarian 45393020 chrl9:59066340- 0.321 Breast, Lung NSC, Ovarian 59471027 chr2: 15977811- 0.056 Lung SC 16073001 chr20:29526118- 0.246 n 29834552 chr20:51603033- 0.371 Hepatocellular, Lung NSC, Ovarian 51989829 chr20:61329497- 0.935 Hepatocellular, Lung NSC 62435964 chr22: 19172385- 0.487 Colorectal, Melanoma, Ovarian 19746441 chrX: 152729030- 1.748 Breast, Lung NSC, Renal 154913754 chrX:66436234- 0.267 Ovarian, Prostate 67090514 In certain embodiments in combination with the amplifications described above (herein), or separately, the CNVs identified as indicative of the presence of a cancer or an increased risk for a cancer include one or more of the deletions shown in Table 5.
TABLE 5. Illustrative, but non-limiting chromosomal segments characterized by ons that are associated with cancers.
Cancer types listed are those identified in him et al. Nature 18: 463: 899-905.
Peak region Length (Mb) Cancer types identified in this is but not prior publications chrl: 388- lpl3.2 Acute lymphoblastic leukemia, Esophageal 119426489 adenocarcinoma, Lung NSC, Lung SC, Melanoma, Ovarian, Prostate chrl :223876038- lq43 Acute lymphoblastic leukemia, Breast, 247249719 Lung SC, ma, Prostate chrl :26377344- 1p36.11 Breast, Esophageal adenocarcinoma, 27532551 Esophageal squamous, Lung NSC, Lung SC, Medulloblastoma, Myeloproliferative 2016/067886 disorder, n, Prostate chrl :3756302- lp36.3 l Acute lymphoblastic leukemia, Breast, 0 Esophageal squamous, Hepatocellular, Lung NSC, Lung SC, Medulloblastoma, Myeloproliferative disorder, Ovarian, Prostate, Renal chrl :71284749- 1p3 l. l Breast, geal adenocarcinoma, 74440273 Glioma, Hepatocellular, Lung NSC, Lung SC, Melanoma, Ovarian, Renal chr2: 1-15244284 2p25.3 Lung NSC, Ovarian chr2: 138479322- 2q22.l , Colorectal, Esophageal 143365272 adenocarcinoma, Esophageal squamous, Hepatocellular, Lung NSC, Ovarian, Prostate, Renal chr2:204533830- 2q33.2 Esophageal adenocarcinoma, 206266883 Hepatocellular, Lung NSC, Medulloblastoma, Renal chr2:24 l 4776 l 9- 2q37.3 Breast, Dedifferentiated liposarcoma, 242951149 Esophageal adenocarcinoma, Esophageal squamous, Hepatocellular, Lung NSC, Lung SC, Medulloblastoma, Melanoma, Ovarian, Renal chr3: 116900556- 3ql3.31 Dedifferentiated liposarcoma, geal 120107320 adenocarcinoma, cellular, Lung NSC, Melanoma, Myeloproliferative disorder, Prostate chr3: 1-2121282 3p26.3 Colorectal, Dedifferentiated liposarcoma, Esophageal adenocarcinoma, Lung NSC, Melanoma, Myeloproliferative disorder chr3: 175446835- 3q26.3 l Acute blastic leukemia, 192 Dedifferentiated liposarcoma, Esophageal adenocarcinoma, Lung NSC, Melanoma, Myeloproliferative disorder, Prostate chr3 :58626894- 3pl4.2 Breast, Colorectal, Dedifferentiated 61524607 liposarcoma, Esophageal adenocarcinoma, Esophageal squamous, Hepatocellular, Lung NSC, Lung SC, Medulloblastoma, Melanoma, roliferative disorder, Ovarian, Prostate, Renal chr4: 93 4pl6.3 roliferative disorder chr4: 186684565- 4q35.2 Breast, Esophageal adenocarcinoma, 191273063 Esophageal squamous, Lung NSC, Medulloblastoma, Melanoma, Prostate, Renal chr4:91089383- 4q22.l Acute lymphoblastic leukemia, Esophageal 93486891 adenocarcinoma, Hepatocellular, Lung NSC, Renal chr5: 177541057- 5q35.3 Breast, Lung NSC, Myeloproliferative 180857866 disorder, Ovarian WO 36059 chr5:57754754- 5qll.2 Breast, Colorectal, Dedifferentiated 59053198 liposarcoma, Esophageal adenocarcinoma, Esophageal squamous, Lung SC, Melanoma, Myeloproliferative disorder, Ovarian, Prostate chr5:85837489- 5q21. l Colorectal, Dedifferentiated liposarcoma, 133480433 Lung NSC, Lung SC, Myeloproliferative disorder, Ovarian chr6: 101000242- 6q22.l Colorectal, Lung NSC, Lung SC 121511318 chr6: 1543157- 6p25.3 ctal, erentiated liposarcoma, 2570302 Esophageal adenocarcinoma, Lung NSC, Lung SC, Ovarian, Prostate chr6: 161612277- 6q26 Colorectal, Esophageal adenocarcinoma, 163134099 Esophageal squamous, Lung NSC, Lung SC, Ovarian, Prostate chr6:76630464- 6ql6. l Colorectal, Hepatocellular, Lung NSC 105342994 chr7: 141592807- 7q34 Breast, Colorectal, Esophageal 142264966 adenocarcinoma, Esophageal squamous, cellular, Lung NSC, n, Prostate, Renal chr7: 144118814- 7q35 Breast, Esophageal adenocarcinoma, 271 Esophageal squamous, Lung NSC, Melanoma, roliferative disorder, Ovarian chr7: 156893473- 7q36.3 Breast, geal adenocarcinoma, 158821424 Esophageal squamous, Lung NSC, Melanoma, Myeloproliferative disorder, Ovarian, Prostate chr7:3046420- 7p22.2 Melanoma, Myeloproliferative disorder, 4279470 n chr7:65877239- 7q21. ll Breast, Medulloblastoma, Melanoma, 79629882 Myeloproliferative disorder, Ovarian chr8: 1-392555 8p23.3 Acute lymphoblastic leukemia, Breast, Myeloproliferative disorder chr8 :2053441- 8p23.2 Acute lymphoblastic leukemia, 6259545 Dedifferentiated liposarcoma, geal arcinoma, Esophageal squamous, Hepatocellular, Lung NSC, Myeloproliferative disorder chr8:22125332- 8p21.2 Acute lymphoblastic leukemia, 30139123 erentiated liposarcoma, Hepatocellular, Myeloproliferative disorder, Ovarian, Renal chr8: 39008109- 8pl 1.22 Acute lymphoblastic leukemia, Breast, 41238710 Dedifferentiated rcoma, Esophageal squamous, Hepatocellular, Lung NSC, Myeloproliferative disorder, Renal chr8 :42971602- 8ql 1.22 Breast, Dedifferentiated liposarcoma, 72924037 Esophageal squamous, Hepatocellular, Lung NSC, Myeloproliferative disorder, Renal chr9:1-708871 9p24.3 Acute blastic leukemia, Breast, Lung NSC, Myeloproliferative disorder, Ovarian, Prostate 1489625- 9p21.3 Colorectal, Esophageal adenocarcinoma, 22474701 Esophageal squamous, roliferative disorder, Ovarian chr9:36365710- 9pl3.2 Myeloproliferative disorder 37139941 chr9: 7161607- 9p24. l Acute lymphoblastic leukemia, Breast, 12713130 Colorectal, geal adenocarcinoma, Hepatocellular, Lung SC, Medulloblastoma, Melanoma, roliferative disorder, Ovarian, Prostate, Renal chrlO: 1-1042949 10pl5.3 Colorectal, Lung NSC, Lung SC, n, Prostate, Renal chrl0: 129812260- 10q26.3 Breast, ctal, Glioma, Lung NSC, 135374737 Lung SC, ma, Ovarian, Renal chrl0:52313829- lOql 1.23 Colorectal, Lung NSC, Lung SC, Ovarian, 53768264 Renal chrl0:89467202- 10q23.3 l Breast, Lung SC, Ovarian, Renal 90419015 chrll:107086196- llq23. l Esophageal adenocarcinoma, 885 Medulloblastoma, Renal chrl 1: 1-1391954 l lpl5.5 Breast, Dedifferentiated liposarcoma, Esophageal adenocarcinoma, Lung NSC, Medulloblastoma, Ovarian chrl 1: 130280899- l lq25 Esophageal adenocarcinoma, Esophageal 134452384 squamous, cellular, Lung NSC, Medulloblastoma, Renal chrll :82612034- llql4.l Melanoma, Renal 85091467 chrl2: 11410696- 12pl3.2 Breast, Hepatocellular, Myeloproliferative 12118386 disorder, Prostate chrl2: 131913408- 33 Dedifferentiated liposarcoma, Lung NSC, 132349534 Myeloproliferative disorder chrl2:9755 l l 77- 12q23. l Breast, Colorectal, Esophageal squamous, 99047626 Lung NSC, Myeloproliferative disorder chr13: 111767404- 13q34 Breast, Hepatocellular, Lung NSC 114142980 chr13: 1-23902184 13q 12.11 Breast, Lung SC, Ovarian chr13 :46362859- 13ql4.2 Hepatocellular, Lung SC, 48209064 Myeloproliferative er, te chr13:9230891 l- 13q3 l.3 Breast, Hepatocellular, Lung NSC, Renal 94031607 chrl4: 1-29140968 14ql 1.2 Acute lymphoblastic leukemia, geal adenocarcinoma, Myeloproliferative disorder chrl4:65275722- 14q23.3 Dedifferentiated liposarcoma, 67085224 Myeloproliferative er chrl4:80741860- 14q32.12 Acute lymphoblastic leukemia, 585 Dedifferentiated liposarcoma, Melanoma, Myeloproliferative disorder chrl5: 1-24740084 15qll.2 Acute lymphoblastic leukemia, Breast, Esophageal adenocarcinoma, Lung NSC, Myeloproliferative disorder, Ovarian chrl5:35140533- 15ql5.l Esophageal adenocarcinoma, Lung NSC, 43473382 Myeloproliferative disorder chrl6: 1-359092 3 Esophageal adenocarcinoma, Hepatocellular, Lung NSC, Renal chrl6:31854743- 16qll.2 Breast, Hepatocellular, Lung NSC, 53525739 Melanoma, Renal chrl6:5062786- 16pl3.3 Hepatocellular, Lung NSC, 7709383 oblastoma, Melanoma, Myeloproliferative disorder, Ovarian, Renal chrl6:76685816- 16q23. l Breast, Colorectal, Esophageal 78205652 adenocarcinoma, Hepatocellular, Lung NSC, Lung SC, Medulloblastoma, Renal 80759878- 16q23.3 Colorectal, Hepatocellular, Renal chrl6:8843693 l- 16q24.3 Colorectal, Hepatocellular, Lung NSC, 88827254 Prostate, Renal chrl7: 10675416- 17pl2 Lung NSC, Lung SC, Myeloproliferative 79 disorder chrl7:26185485- l 7q11.2 Breast, Colorectal, Dedifferentiated 27216066 liposarcoma, Lung NSC, Lung SC, Melanoma, Myeloproliferative disorder, Ovarian chrl7:37319013- l 7q21.2 Breast, Colorectal, Dedifferentiated 37988602 liposarcoma, Lung SC, Melanoma, Myeloproliferative disorder, Ovarian chrl7:7471230- 17pl3.l Lung SC, Myeloproliferative disorder 7717938 78087533- l 7q25.3 Colorectal, Myeloproliferative disorder 78774742 chrl8: 1-587750 18pl 1.32 Myeloproliferative disorder 46172638- 18q21.2 Esophageal adenocarcinoma, Lung NSC 49935241 chrl8:75796373- 18q23 Colorectal, geal adenocarcinoma, 76117153 Esophageal squamous, Ovarian, Prostate chrl9: 1-526082 19pl3.3 Hepatocellular, Lung NSC, Renal chrl9:21788507- 19pl2 Hepatocellular, Lung NSC, Renal 34401877 52031294- 19ql3.32 Breast, Hepatocellular, Lung NSC, 53331283 Medulloblastoma, Ovarian, Renal chrl9:63402921- 19ql3.43 Breast, Colorectal, Dedifferentiated 63811651 liposarcoma, Hepatocellular, Lung NSC, Medulloblastoma, Ovarian, Renal chr20: 1-325978 20pl3 Breast, Dedifferentiated liposarcoma, Lung chr20: 14210829- 20pl2.l Esophageal adenocarcinoma, Lung NSC, 15988895 Medulloblastoma, Melanoma, Myeloproliferative disorder, Prostate, Renal chr21 860- 2lq22.2 Breast 42033506 chr22:205 l 766 l- 22ql 1.22 Acute blastic ia, Esophageal 23 adenocarcinoma chr22:45488286- 22ql3.33 Breast, Hepatocellular, Lung NSC, Lung 49691432 SC chrX: 1-3243111 Xp22.33 Esophageal adenocarcinoma, Lung NSC, Lung SC chrX:31041721- Xp21.2 Acute lymphoblastic leukemia, geal 34564697 arcinoma, Glioma ] The anuploidies fied as characteristic ofvarious cancers (e.g., the anuploidies identified in Tables 4 and 5) may contain genes known to be implicated in cancer etiologies (e.g., tumor suppressors, oncogenes, etc.). These aneuploidies can also be probed to identifiy relevant but previously unknown genes.
[00469] For example Beroukhim et al. supra, assessed potential cancer-causing genes in the copy number alterations using GRAIL (Gene Relationships Among Implicated Loci20), an algorithm that searches for functional relationships among genomic regions. GRAIL scores each gene in a collection of genomic s for its 'relatedness'to genes in other regions based on textual similarity between published abstracts for all papers citing the genes, on the notion that some target genes will function in common pathways. These methods permit identification/characterization of genes usly not associated with the particular cancers at issue. Table 6 illustrates target genes known to be within the identified amplified t and predicted genes, and Table 7 illustrates target genes known to be within the identified deleted segment and predicted genes. 2016/067886 TABLE 6. Illustrative, but non-limiting chromosomal segments and genes known or predicted to be present in regions characterized by amplification in various cancers (see, e.g.• him et al. supra.).
Chromosome Peak region #genes Known GRAIL top and band tareet tareet 8q24.21 chr8: 432- 1 MYC MYC 128849112 llql3.2 chrl 1:69098089- 3 CCNDI ORAOVI 69278404 17ql2 chrl7:35067383- 6 ERBB2 ERBB2, 35272328 CJ7orf37 1 chrl2:56419524- 7 CDK4 TSPAN31 56488685 14ql3.3 chrl4:35708407- 3 NKX2-I NKX2-I 36097605 12ql5 chrl2:67440273- 1 MDM2 MDM2 67566002 7pl 1.2 chr7:54899301- 1 EGFR EGFR 55275419 lq21.2 chrl: 148661965- 9 MCLI MCLI 149063439 8pl2 chr8:38252951- 3 FGFRI FGFRI 38460772 l chrl2:25 l 89655- 2 KRAS KRAS 25352305 19ql2 3497553 l- 1 CCNEI CCNEI 35098303 22ql 1.21 chr22: 19172385- 11 CRKL CRKL 19746441 12ql5 chrl2:68249634- 2 LRRCIO 68327233 12ql4.3 chr12: 64461446- 1 HMGA2 HMGA2 64607139 Xq28 chrX: 152729030- 53 SPRY3 154913754 5pl5.33 chr5: 1212750- 3 TERT TERT 1378766 3q26.2 chr3: 170024984- 22 PRKCI PRKCI 173604597 15q26.3 chrl5:96891354- 4 IGFIR IGFIR 97698742 20ql3.2 chr20:51603033- 1 ZNF217 51989829 8pl 1.21 chr8:42006632- 6 PLAT 42404492 lp34.2 chrl :39907605- 7 MYCLI MYCLI 40263248 2016/067886 l 7q21.33 chrl7:44673157- 4 NGFR,PHB 45060263 2p24.3 chr2: 11- 1 MYCN MYCN 16073001 7q21.3 chr7:89924533- 62 CDK6 CDK6 98997268 13q34 chr13: 108477140- 4 IRS2 110084607 l chrl 1:76699529- 14 GAB2 78005085 20ql3.33 chr20:61329497- 38 BIRC7 62435964 l 7q23. l chrl7:55144989- 5 I 55540417 lpl2 chrl: 566- 5 REG4 120303234 8q21.13 chr8:81242335- 3 ZNF704, 81979194 ZBTBIO 6p21. l chr6:43556800- 18 VEGFA 44361368 5pll chr5:45312870- 0 49697231 20qll.21 chr20:29526118- 5 BCL2LI BCL2LI, !DI 29834552 6q23.3 chr6: 135561194- 1 MYB hsa-mir-548a- 135665525 2 lq44 chrl :241364021- 71 AKT3 247249719 5q35.3 chr5: 174477192- 92 FLT4 180857866 7q3 l.2 chr7:115981465- 3 MET MET 116676953 18qll.2 chrl8: 17749667- 21 CABLESI 22797232 l 7q25. l chrl7:70767943- 13 GRB2, ITGB4 71305641 lp32. l chrl :58658784- 7 JUN JUN 60221344 17qll.2 chrl7:24112056- 5 DHRS13, 24310787 FLOT2, ERALI, PHF12 l 7pl 1.2 chrl7: 18837023- 12 MAPK7 19933105 8q24.11 chr8: 116186189- 13 NOV 120600761 12ql5 66458200- 0 66543552 19ql3.2 chrl9:43 l 77306- 60 LGALS7, 45393020 DYRKIB llq22.2 chrll:101433436- 8 BIRC2, BIRC2 102134907 YAPI 4ql2 chr4: 54471680- 7 PDGFRA, KDR,KIT 55980061 KIT 12pl 1.21 chr12: 30999223- 9 DDXll, 32594050 FAM60A 3q28 chr3: 178149984- 143 PIK3CA PIK3CA 199501827 lp36.33 chrl: 1-5160566 77 TP73 l 7q24.2 chrl7:62318152- 12 BPTF 63890591 lq23.3 chrl: 017- 52 PEAl5 159953843 lq24.3 chrl: 169549478- 6 BAT2DI, 170484405 MYOC 8q22.3 chr8: 101163387- 14 RRM2B 103693879 13q3 l.3 chr13 :89500014- 3 GPC5 93206506 12q21.l chrl2:70849987- 0 70966467 12pl3.33 chr12: 1-1311104 10 WNKI 12q21.2 chrl2:76852527- 0 lq32. l chrl :201678483- 21 MDM4 MDM4 203358272 19ql3.42 chrl9:59066340- 19 PRKCG, 59471027 TSEN34 12ql2 38788913- 12 ADAMTS20 42596599 12q23. l chrl2:95089777- 2 ELK3 95350380 12q21.32 85072329- 0 85674601 10q22.3 chrl0:74560456- 46 SFTPAIB 82020637 3pl 1.1 chr3: 86250885- 8 POUIFI 95164178 17qll.l chrl7:22479313- I WSBI 8q24.3 chr8:140458177- 97 PTP4A3, 826 MAFA, PARPIO Xql2 chrX:66436234- I AR AR 67090514 6ql2 chr6:63255006- 3 PTP4AI 65243766 14qll.2 chrl4: 1-23145193 95 BCL2L2 9q34.3 chr9: 137859478- 76 NRARP, 140273252 MRPL41, TRAF2, LHX3 6p24. l chr6: 1-23628840 95 E2F3 13ql2.2 chr13: 1-40829685 110 FOXOI l chrl2:72596017- 0 14q32.33 chr14: 106074644- 0 106368585 llpl3 chrll :32027116- 35 WTI 37799354 TABLE 7. Illustrative, but non-limiting chromosomal segments and genes known or predicted to be present in regions charactierzed by amplification in various cancers (see, e.g.• Beroukhim et al. supra.).
Chromoso Peak region # Known GRAIL meand gene target top target band s 9p21.3 chr9:21489625- 5 CDKN2A CDKN2A 22474701 IE 3pl4.2 chr3 :58626894- 2 FHIT FHIT 16q23. l chrl6:76685816- 2 wwox wwox 78205652 9p24. l chr9:7161607- 3 PTPRD PTPRD 12713130 20pl2. l chr20: 14210829- 2 MACRO FLRT3 15988895 D2 6q26 chr6: 161612277- 1 PARK2 PARK2 163134099 13ql4.2 chr13 :46362859- 8 RBI RBI 2q22. l chr2: 138479322- 3 LRPIB LRPIB 143365272 4q35.2 chr4: 186684565- 15 FRG2, 191273063 TUBB4Q 5ql 1.2 7754754- 5 PDE4D PLK2, 59053198 PDE4D 16pl3.3 chrl6:5062786- 2 A2BPI A2BPI 7709383 7q34 chr7: 141592807- 3 TRB PRSSI 142264966 2q37.3 chr2:24 l 4776 l 9- 19 G 242951149 ,ING5 19pl3.3 chrl9: 1-526082 10 GZA1M, THEG, PPAP2C, CJ9orf20 10q23.3 l 89467202- 4 PTEN PTEN 90419015 8p23.2 chr8:2053441- 1 CSMDI CSMDI 6259545 lp36.3 l chrl :3756302- 23 DFFB, 6867390 ZBTB48, AJAPI 4q22. l chr4:91089383- 2 MGC4862 93486891 8 18q23 chrl8:75796373- 4 PARD6G 76117153 6p25.3 chr6: 1543157- 2 FOXCI 2570302 19ql3.43 chrl9:63402921- 17 ZNF324 63811651 Xp21.2 chrX:31041721- 2 DMD DMD 34564697 l lq25 chrl 1: 130280899- 12 OPCML, HNT 134452384 HNT 13q 12.11 chr13: 1-23902184 29 LATS2 22ql3.33 chr22:45488286- 38 TUBGCP6 49691432 15qll.2 chrl5: 0084 20 A26BI 22ql 1.22 chr22:205 l 766 l- 3 VPREBI 21169423 10q26.3 chrl0: 129812260- 35 MGMT, 135374737 SYCEI 2 chrl2: 11410696- 2 ETV6 ETV6 12118386 8p23.3 chr8: 1-392555 2 ZNF596 1p36.11 chrl :26377344- 24 SFN 27532551 l lpl5.5 chrl 1: 1-1391954 49 RASSF7 2 chrl7:26185485- 10 NFI NFI 27216066 llq23. l 107086196- 61 ATM CADMI 116175885 9p24.3 chr9:1-708871 5 FOXD4 lOql 1.23 chrl0:52313829- 4 PRKGI DKKI, 53768264 PRKGI 15ql5.l chrl5:35140533- 109 4 43473382 lpl3.2 chrl: 110339388- 81 MAGI3 119426489 Xp22.33 chrX: 1-3243111 21 SHOX 3p26.3 chr3: 1-2121282 2 CHLI 9pl3.2 chr9:36365710- 2 PAX5 MELK 37139941 17pl3.l chrl7:7471230- 10 TP53 ATPIB2 7717938 12q24.33 chrl2: 131913408- 7 CHFR 132349534 7q36.3 chr7: 156893473- 7 PTPRN2 NCAPG2 158821424 6ql6. l chr6:76630464- 76 FUT9, 105342994 C6orfl65, C6orfl62, GJAIO 5q21. l chr5:85837489- 142 AFC AFC 133480433 8pl 1.22 chr8:39008109- 7 C8orf4, ZMAT4 19ql3.32 chrl9:52031294- 25 BBC3 53331283 10pl5.3 chrlO: 1-1042949 4 TUBB8 1p3 l. l chrl :71284749- 4 NEGRI NEGRI 74440273 13q3 l.3 9230891 l- 2 GPC6 GPC6, 94031607 DCT 16qll.2 31854743- 37 RBL2 53525739 20pl3 chr20: 1-325978 10 SOXl2 5q35.3 chr5: 177541057- 43 SCGB3AI 180857866 lq43 chrl :223876038- 173 RYR2 FH, 247249719 ZNF678 16pl3.3 chrl6: 1-359092 16 HBZ l 7q21.2 chrl7:37319013- 22 CNP 37988602 2p25.3 chr2: 1-15244284 51 MYTIL 3ql3.31 chr3: 556- 1 LSAMP 120107320 7q21. ll chr7:65877239- 73 MAGI2 CLDN4 79629882 7q35 chr7: 144118814- 3 CNTNAP CNTNAP2 271 2 14q32.12 chrl4:80741860- 154 PRIMAi 106368585 16q24.3 chrl6:8843693 l- 9 Cl6orf3 88827254 3q26.3 l chr3: 175446835- 1 NAALAD NMLADL 192 L2 2 l 7q25.3 chrl7:78087533- 8 ZNF750 78774742 19pl2 chrl9:21788507- 12 ZNF492, 34401877 ZNF99 12q23. l 9755 l l 77- 3 ANKSIB ANKSIB 4pl6.3 chr4: 1-435793 4 ZNF141 18pl 1.32 chrl8: 1-587750 4 COLEC12 2q33.2 chr2:204533830- 1 PARD3B PARD3B 206266883 8p21.2 chr8:22125332- 63 DPYSL2, 30139123 STMN4 8ql 1.22 chr8 :42971602- 86 SNTGI FLJ23356, 72924037 STl8, RBICCI 16q23.3 chrl6:80759878- 2 CDHl3 CDHl3 82408573 llql4.l chrll :82612034- 6 DLG2 CCDC89, 85091467 CCDC90B, TMEMl26 14q23.3 chrl4:65275722- 7 GPHN, 67085224 MPP5 7p22.2 chr7:3046420- 1 SDKI SDKI 4279470 13q34 chr13: 111767404- 25 TUBGCP3 114142980 17pl2 chrl7: 10675416- 5 MAP2K4 MAP2K4, 12635879 ZNF18 2lq22.2 chr21 :38584860- 19 DSCAM, DSCAM 42033506 TMPRSS 2/ERG 18q21.2 chrl8:46172638- 7 SMAD4, DCC 41 DCC 6q22. l chr6: 101000242- 87 GTF3C6, 121511318 TUBEI, ROSI 14qll.2 chrl4: 1-29140968 140 ZNF219, NDRG2 In vanous embodiments, it is plated to use the methods identified herein to identify CNVs of segments comprising the amplified regions or genes fied in Table 6 and/or to use the methods fied herein to identify CNVs of segments comprising the deleted regions or genes identified in 7.
[00471] In one embodiment, the methods described herein provide a means to assess the association n gene amplification and the extent of tumor evolution.
Correlation between amplification and/or on and stage or grade of a cancer may be prognostically important because such information may contribute to the definition of a genetically based tumor grade that would better predict the future course of disease with more advanced tumors having the worst prognosis. In addition, information about early amplification and/or deletion events may be useful in ating those events as predictors of subsequent disease progression.
[00472] Gene amplification and deletions as identified by the method can be associated with other known parameters such as tumor grade, histology, d labeling index, hormonal status, nodal involvement, tumor size, survival duration and other tumor properties available from epidemiological and biostatistical s. For example, tumor DNA to be tested by the method could include atypical hyperplasia, ductal carcinoma in situ, stage I-III cancer and metastatic lymph nodes in order to permit the identification of associations between ications and deletions and stage. The associations made may make possible effective therapeutic intervention.
For e, consistently amplified regions may contain an overexpressed gene, the product of which may be able to be attacked therapeutically (for e, the growth factor receptor tyrosine kinase, pl85HER2).
In various embodiments, the methods described herein can be used to identify amplification and/or deletion events that are associated with drug resistance by determining the copy number variation of nucleic acid sequences from primary cancers to those of cells that have metastasized to other sites. If gene amplification and/or deletion is a manifestation of karyotypic instability that allows rapid development of drug resistance, more amplification and/or deletion in primary tumors from chemoresistant patients than in tumors in chemosensitive patients would be expected. For example, if amplification of specific genes is sible for the development of drug resistance, regions surrounding those genes would be expected to be amplified consistently in tumor cells from pleural effusions of esistant patients but not in the primary tumors. Discovery of associations between gene amplification and/or deletion and the development of drug resistance may allow the fication of patients that will or will not t from adjuvant therapy.
In a manner similar to that described for determining the presence or absence of complete and/or partial fetal chromosomal aneuploidies in a maternal sample, methods, tus, and systems described herein can be used to determine the presence or absence of te and/or partial somal aneuploidies in any patient sample comprising c acids e.g. DNA or cIDNA (including patient WO 36059 samples that are not maternal samples). The patient sample can be any biological sample type as described elsewhere herein. Preferably, the sample is obtained by non-invasive procedures. For example, the sample can be a blood sample, or the serum and plasma fractions f. Alternatively, the sample can be a urine sample or a fecal sample. In yet other embodiments, the sample is a tissue biopsy sample. In all cases, the sample comprises nucleic acids e.g. cfDNA or genomic DNA, which is purified, and ced using any of the NGS sequencing methods described previously.
Both complete and partial chromosomal aneuploidies associated with the formation, and progression of cancer can be determined according to the present method.
] In various embodiments, when using the methods bed herein to determine the presence and/or increased risk of cancer normalization of the data can be made with respect to the some(s) for which the CNV is determined. In certain embodiments normalization of the data can be made with respect to the chromosome arm(s) for which the CNV is ined. In certain embodiments, normalization of the data can be made with respect to the particular segment(s) for which the CNV is determined.
In addition to the role of CNV in cancer, CNVs have been associated with a growmg number of common complex disease, including human immunodeficiency virus (HIV), autoimmune diseases and a um of neuropsychiatric disorders.
CNVs in ious and autoimmune disease To date a number of studies have reported association between CNV in genes involved in inflammation and the immune se and HIV, asthma, Crohn's disease and other autoimmune disorders (Fanciulli et al., Clin Genet 77:201-213 ). For example, CNV in CCL3LI, has been implicated in HIV/AIDS susceptibility (CCL3LI, l 7q11.2 deletion), toid arthritis (CCL3LI, l 7q11.2 deletion), and Kawasaki disease (CCL3LI, l 7q11.2 duplication); CNV in HBD-2, has been reported to predispose to colonic Crohn's disease (HDB-2, 8p23. l deletion) and psoriasis , 8p23. l deletion); CNV in FCGR3B, was shown to predispose to ulonephritis in systemic lupus erthematosous (FCGR3B, lq23 deletion, lq23 duplication), anti-neutrophil cytoplasmic antibody (ANCA)-associated vasculatis (FCGR3B, lq23 deletion), and increase the risk of developing rheumatoid arthritis.
There are at least two matory or autoimmune diseases that have been shown to be associated with CNV at different gene loci. For example, Crohn's disease is associated with low copy number at HDB-2, but also with a common deletion polymorphism upstream of the IGRM gene that s a member of the p47 immunity-related GTPase family. In addition to the association with FCGR3B copy number, SLE susceptibility has also been reported to be significantly sed among subjects with a lower number of copies of complement component C4.
[00479] ations between genomic deletions at the GSTMI , letion) and GSTTI (GSTTI, 22q11.2 deletion) loci and increased risk of atopic asthma have been reported in a number of independent studies. In some ments, the methods described herein can be used to determine the presence or absence of a CNV ated with inflammation and/or autoimmune es. For example, the methods can be used to determine the presence of a CNV in a patient suspected to be suffering from HIV, asthma, or Crohn's disease. Examples of CNV associated with such diseases include without limitation deletions at 17q11.2, 8p23 .1, lq23, and 22ql 1.2, and duplications at l 7ql 1.2, and lq23. In some embodiments, the present method can be used to determine the presence of CNV in genes including but not limited to CCL3LI, HBD-2, FCGR3B, GSTM, GSTTI, C4, and IRGM.
CNVdiseases ofthe s system ations between de nova and inherited CNV and several common neurological and psychiatric diseases have been ed in autism, phrenia and epilepsy, and some cases of neurodegenerative diseases such as Parkinson's disease, amyotrophic lateral sclerosis (ALS) and autosomal dominant Alzheimer's disease (Fanciulli et al., Clin Genet 77:201-213 [2010]). Cytogenetic abnormalities have been observed in patients with autism and autism spectrum disorders (ASDs) with duplications at l 5q11-q13. According to the Autism Genome project Consortium, 154 CNV including several recurrent CNVs, either on chromosome 15ql l-ql3 or at new genomic locations including chromosome 2pl6, lq21 and at l 7pl2 in a region associated with Smith-Magenis syndrome that overlaps with ASD. Recurrent microdeletions or microduplications on chromosome 16pl 1.2 have highlighted the ation that de nova CNVs are detected at loci for genes such as SHANK3 .3 deletion), neurexin 1 , 2pl6.3 deletion) and the neuroglins (NLGN4, Xp22.33 deletion) that are known to regulate synaptic differentiation and regulate inergic neurotransmitter release. Schizophrenia has also been associated with multiple de nova CNVs. Microdeletions and microduplications associated with phrenia n an overrepresentation of genes belonging to neurodevelopmental and glutaminergic pathways, suggesting that multiple CNVs ing these genes may contribute directly to the pathogenesis of schizophrenia e.g. ERBB4, 2q34 deletion, SLCIA3, 5pl3.3 deletion; RAPEGF4, 2q3 l. l deletion; CIT, 12.24 deletion; and multiple genes with de nova CNV. CNVs have also been associated with other neurological disorders including epilepsy (CHRNA7, 15ql3.3 deletion), Parkinson's disease (SNCA 4q22 duplication) and ALS (SMNI, 5ql2.2.-ql3.3 deletion; and SMN2 on). In some embodiments, the methods described herein can be used to determine the presence or absence of a CNV associated with es of the nervous system. For example, the methods can be used to determine the presence of a CNV in a patient suspected to be suffering from autisim, phrenia, epilepsy, egenerative diseases such as Parkinson's disease, ophic lateral sclerosis (ALS) or autosomal dominant Alzheimer's disease. The methods can be used to determine CNV of genes associated with diseases of the nervous system ing without limitation any of the Autism Spectrum Disorders (ASD), schizophrenia, and epilepsy, and CNV of genes associated with neurodegenerative disorders such as Parkinson's disease. Examples of CNV ated with such diseases include without limitation duplications at 15qll-ql3, 2pl6, lq21, l 7pl2, 16pl 1.2, and 4q22, and deletions at 22ql3.3, , Xp22.33, 2q34, 5pl3.3, 2q31.l, 12.24, 15ql3.3, and 5ql2.2. In some embodiments, the methods can be used to determine the presence of CNV in genes including but not limited to SHANK3, NLGN4, NRXNI, ERBB4, SLCIA3, RAPGEF4, CIT, CHRNA7, SNCA, SMNl,and SMN2.
CNVand metabolic or cardiovascular diseases
[00481] The ation between metabolic and cardiovascular traits, such as familial hypercholesterolemia (FH), atherosclerosis and coronary artery disease, and CNVs has been reported in a number of studies (Fanciulli et al., Clin Genet 77:201- 213 [2010]). For example, germline rearrangements, mainly deletions, have been WO 36059 observed at the LDLR gene (LDLR, 19pl3.2 deletion/duplication) in some FH patients who carry no other LDLR mutations. Another example is the LPA gene that encodes apolipoprotein(a) (apo(a)) whose plasma concentration is associated with risk of coronary artery disease, myocardial tion (MI) and stroke. Plasma concentrations ofthe apo(a) containing lipoprotein Lp(a) vary over 1000-fold n individuals and 90% of this variability is genetically determined at the LPA locus, with plasma concentration and Lp(a) isoform size being proportional to a highly variable number of 'kringle 4' repeat sequences (range 5-50). These data indicate that CNV in at least two genes can be associated with vascular risk. The methods described herein can be used in large studies to search specifically for CNV associations with cardiovascular disorders. In some embodiments, the present method can be used to determine the presence or absence of a CNV associated with metabolic or vascular disease. For example, the present method can be used to determine the presence of a CNV in a patient ted to be suffering from familial hypercholesterolemia. The methods described herein can be used to determine CNV of genes associated with metabolic or cardiovascular disease e.g. hypercholesterolemia. Examples of CNV associated with such diseases include without limitation 19pl3.2 deletion/duplication of the LDLR gene, and multiplications in the LPA gene.
Apparatus and systems for determining CNV Analysis of the sequencing data and the diagnosis d therefrom are typically med using various computer executed thms and programs.
Therefore, certain embodiments employ processes involving data stored in or erred h one or more computer systems or other sing systems.
Embodiments disclosed herein also relate to tus for performing these operations. This apparatus may be specially ucted for the required purposes, or it may be a general-purpose computer (or a group of computers) selectively activated or reconfigured by a computer program and/or data structure stored in the computer.
In some embodiments, a group of processors performs some or all of the recited analytical operations collaboratively (e.g., via a network or cloud computing) and/or in parallel. A processor or group of processors for performing the methods described herein may be of various types including microcontrollers and microprocessors such as programmable devices (e.g., CPLDs and FPGAs) and non-programmable s such as gate array ASICs or l purpose microprocessors.
In addition, certain embodiments relate to tangible and/or nontransitory computer readable media or computer program products that include program ctions and/or data (including data structures) for performing various computer-implemented operations. Examples of er-readable media e, but are not limited to, semiconductor memory devices, magnetic media such as disk drives, magnetic tape, l media such as CDs, magneto-optical media, and re devices that are specially configured to store and perform program instructions, such as read-only memory devices (ROM) and random access memory (RAM). The er readable media may be directly controlled by an end user or the media may be indirectly controlled by the end user. Examples of directly controlled media include the media located at a user facility and/or media that are not shared with other entities. Examples of indirectly controlled media include media that is ctly accessible to the user via an al network and/or via a service providing shared resources such as the "cloud." Examples of program instructions include both machine code, such as produced by a compiler, and files containing higher level code that may be executed by the computer using an interpreter.
In various embodiments, the data or information ed m the disclosed methods and apparatus is ed in an electronic format. Such data or information may include reads and tags d from a nucleic acid sample, counts or ies of such tags that align with particular regions of a reference sequence (e.g., that align to a chromosome or chromosome segment), reference sequences (including reference sequences providing solely or primarily polymorphisms), chromosome and segment doses, calls such as aneuploidy calls, normalized chromosome and segment values, pairs of chromosomes or segments and corresponding normalizing chromosomes or segments, counseling recommendations, diagnoses, and the like. As used herein, data or other information provided in electronic format is available for storage on a machine and transmission between machines. Conventionally, data in electronic format is provided digitally and may be stored as bits and/or bytes in various data structures, lists, databases, etc. The data may be ed el ectronically, optically, etc. 2016/067886 One ment provides a computer program product for generating an output indicating the presence or absence of an aneuploidy, e.g., a fetal aneuploidy or , in a test sample. The computer product may contain instructions for performing any one or more of the described s for determining a chromosomal anomaly. As explained, the computer product may include a nontransitory and/or tangible computer readable medium having a computer executable or able logic (e.g., instructions) recorded thereon for enabling a processor to ine some doses and, in some cases, whether a fetal aneuploidy is present or absent. In one example, the computer product ses a computer le medium having a computer executable or compilable logic (e.g., instructions) recorded thereon for enabling a processor to diagnose a fetal aneuploidy compnsmg: a receiving procedure for ing sequencing data from at least a portion of nucleic acid molecules from a maternal ical sample, wherein said sequencing data comprises a calculated chromosome and/or segment dose; er assisted logic for analyzing a fetal aneuploidy from said received data; and an output procedure for generating an output indicating the presence, absence or kind of said fetal aneuploidy.
The sequence information from the sample under consideration may be mapped to chromosome reference sequences to identify a number of sequence tags for each of any one or more chromosomes of interest and to fy a number of ce tags for a normalizing segment sequence for each of said any one or more chromosomes of interest. In various embodiments, the reference sequences are stored in a database such as a relational or object database, for example.
It should be understood that it is not practical, or even possible in most cases, for an unaided human being to perform the computational operations of the methods disclosed herein. For example, mapping a single 30 bp read from a sample to any one of the human chromosomes might require years of effort without the assistance of a computational apparatus. Of course, the problem is compounded because reliable aneuploidy calls generally require mapping thousands (e.g., at least about 10,000) or even millions ofreads to one or more chromosomes.
The methods disclosed herein can be performed using a system for tion of copy number of a genetic sequence of interest in a test sample. The system comprising: (a) a sequencer for receiving nucleic acids from the test sample providing nucleic acid sequence information from the sample; (b) a processor; and (c) one or more computer-readable storage media having stored thereon instructions for ion on said processor to carry out a method for identifying any CNV, e.g., chromosomal or l oidies.
In some embodiments, the methods are instructed by a computer-readable medium having stored thereon computer-readable instructions for carrying out a method for identifying any CNV, e.g., chromosomal or partial aneuploidies. Thus one embodiment provides a computer program product comprising one or more erreadable non-transitory storage media having stored thereon computer-executable instructions that, when executed by one or more processors of a computer system, cause the computer system to implement a method for tion of copy number of a sequence of interest in a test sample comprising fetal and maternal cell-free nucleic acids. The method includes: (a) receiving sequence reads obtained by sequencing the cell-free nucleic acid fragments in the test sample; (b) aligning the sequence reads of the cell-free nucleic acid fragments to a reference genome comprising the sequence of interest, thereby providing test sequence tags, wherein the reference genome is divided into a ity of bins; (c) determining sizes of the cell-free nucleic acid fragments existing in the test sample; (d) weighting the test sequence tags based on the sizes of cell-free c acid fragments from which the tags are obtained; (e) calculating coverages for the bins based on the weighted tags of (d); and (f) identifying a copy number variation in the sequence of interest from the calculated coverages. In some implementations, weighting the test ce tags involves biasing the coverages toward test sequence tags obtained from cell-free c acid fragments of a size or a size range characteristic of one genome in the test sample. In some implementations, weighting the test sequence tags involves assigning a value of 1 to tags obtained from ree nucleic acid fragments of the size or the size range, and ing a value of 0 to other tags. In some implementations, the method further involves determining, in bins of the reference genome, including the sequence of interest, values of a fragment size parameter including a ty of the cell-free nucleic acid nts in the test sample having fragment sizes r or longer than a threshold value. Here, identifying the copy number variation in the ce of interest involves using the values of the fragment size parameter as well as the coverages calculated in (e). In some implementations, the system is configured to 2016/067886 evaluate copy number in the test sample using the various methods and processes sed above.
In some embodiments, the instructions may further include automatically recording information pertinent to the method such as chromosome doses and the presence or absence of a fetal chromosomal aneuploidy in a patient medical record for a human subject providing the maternal test . The patient medical record may be maintained by, for example, a laboratory, physician's office, a hospital, a health maintenance organization, an insurance y, or a personal medical record website. Further, based on the results of the processor-implemented is, the method may further involve prescribing, initiating, and/or altering treatment of a human subject from whom the maternal test sample was taken. This may involve performing one or more additional tests or analyses on additional samples taken from the subject.
[00490] Disclosed methods can also be performed using a computer processing system which is adapted or ured to perform a method for identifying any CNV, e.g., chromosomal or partial aneuploidies. One embodiment provides a computer processing system which is adapted or configured to perform a method as described herein. In one embodiment, the apparatus comprises a sequencing device adapted or ured for sequencing at least a portion ofthe nucleic acid molecules in a sample to obtain the type of sequence information bed elsewhere herein. The apparatus may also include components for processing the sample. Such components are described ere herein.
Sequence or other data, can be input into a computer or stored on a er readable medium either directly or indirectly. In one embodiment, a computer system is directly coupled to a sequencing device that reads and/or analyzes sequences of nucleic acids from samples. Sequences or other information from such tools are provided via interface in the computer system. Alternatively, the ces processed by system are provided from a ce storage source such as a database or other repository. Once available to the processing apparatus, a memory device or mass storage device buffers or stores, at least temporarily, sequences of the nucleic acids. In addition, the memory device may store tag counts for various chromosomes or genomes, etc. The memory may also store s routines and/or programs for 2016/067886 analyzing the presenting the sequence or mapped data. Such ms/routines may include programs for performing statistical analyses, etc.
In one e, a user provides a sample into a cing apparatus.
Data is collected and/or analyzed by the sequencing tus which is connected to a computer. Software on the computer allows for data collection and/or analysis. Data can be stored, displayed (via a monitor or other similar device), and/or sent to another location. The computer may be connected to the internet which is used to it data to a handheld device utilized by a remote user (e.g., a physician, scientist or t). It is understood that the data can be stored and/or analyzed prior to transmittal. In some embodiments, raw data is collected and sent to a remote user or apparatus that will analyze and/or store the data. Transmittal can occur via the internet, but can also occur via satellite or other connection. ately, data can be stored on a computer-readable medium and the medium can be shipped to an end user (e.g., via mail). The remote user can be in the same or a different geographical location including, but not limited to a building, city, state, country or continent.
In some embodiments, the methods also include collecting data regarding a plurality of polynucleotide sequences (e.g., reads, tags and/or nce chromosome sequences) and sending the data to a computer or other computational system. For example, the computer can be connected to laboratory equipment, e.g., a sample collection apparatus, a nucleotide amplification tus, a nucleotide sequencing apparatus, or a ization apparatus. The computer can then collect applicable data gathered by the laboratory device. The data can be stored on a computer at any step, e.g., while collected in real time, prior to the sending, during or in conjunction with the sending, or following the sending. The data can be stored on a computer-readable medium that can be extracted from the computer. The data collected or stored can be transmitted from the computer to a remote location, e.g., via a local network or a wide area network such as the internet. At the remote location various operations can be performed on the transmitted data as described below.
Among the types of electronically formatted data that may be stored, transmitted, ed, and/or manipulated in systems, apparatus, and methods disclosed herein are the following: Reads obtained by sequencing nucleic acids in a test sample Tags obtained by aligning reads to a reference genome or other reference sequence or ces The nce genome or sequence Sequence tag density - Counts or s of tags for each of two or more regions (typically chromosomes or chromosome segments) of a reference genome or other reference sequences Identities of normalizing chromosomes or chromosome segments for particular chromosomes or chromosome segments ofinterest Doses for chromosomes or chromosome segments (or other regions) obtained from chromosomes or segments of interest and corresponding normalizing somes or segments Thresholds for calling chromosome doses as either affected, non-affected, or no call The actual calls of chromosome doses ses (clinical condition associated with the calls) Recommendations for further tests derived from the calls and/or ses ent and/or monitoring plans derived from the calls and/or diagnoses These various types of data may be obtained, stored transmitted, analyzed, and/or manipulated at one or more locations using distinct apparatus. The processing options span a wide spectrum. At one end of the spectrum, all or much of this information is stored and used at the location where the test sample is processed, e.g., a doctor's office or other clinical g. In other extreme, the sample is obtained at one location, it is processed and optionally sequenced at a different location, reads are aligned and calls are made at one or more different locations, and diagnoses, recommendations, and/or plans are prepared at still another location (which may be a location where the sample was obtained).
In various embodiments, the reads are generated with the sequencing apparatus and then transmitted to a remote site where they are processed to produce aneuploidy calls. At this remote on, as an example, the reads are aligned to a reference sequence to produce tags, which are counted and assigned to chromosomes or segments of interest. Also at the remote location, the counts are ted to doses using associated normalizing chromosomes or segments. Still further, at the remote on, the doses are used to generate aneuploidy calls.
WO 36059 Among the processing operations that may be employed at distinct locations are the following: Sample collection Sample processing preliminary to sequencing Sequencing ing sequence data and deriving aneuploidy calls Diagnosis Reporting a diagnosis and/or a call to patient or health care provider Developing a plan for further treatment, testing, and/or monitoring Executing the plan ling Any one or more of these operations may be automated as bed ere herein. Typically, the sequencing and the analyzing of sequence data and deriving aneuploidy calls will be performed computationally. The other operations may be performed manually or tically.
Examples of locations where sample tion may be performed include health practitioners' offices, clinics, patients' homes (where a sample collection tool or kit is provided), and mobile health care vehicles. Examples of locations where sample processing prior to sequencing may be performed include health practitioners' offices, clinics, patients' homes (where a sample processing apparatus or kit is provided), mobile health care vehicles, and ties of aneuploidy analysis providers. Examples of locations where sequencing may be performed include health practitioners' offices, clinics, health practitioners' offices, clinics, patients' homes (where a sample sequencing apparatus and/or kit is provided), mobile health care vehicles, and ties of aneuploidy analysis providers. The location where the sequencing takes place may be provided with a ted network connection for transmitting sequence data (typically reads) in an electronic format.
Such connection may be wired or wireless and have and may be configured to send the data to a site where the data can be processed and/or ated prior to transmission to a processing site. Data ators can be maintained by health organizations such as Health Maintenance Organizations (HMOs). 2016/067886 The analyzing and/or deriving operations may be performed at any of the foregoing locations or alternatively at a further remote site dedicated to computation and/or the service of analyzing nucleic acid sequence data. Such ons include for example, rs such as general purpose server farms, the facilities of an aneuploidy analysis service business, and the like. In some ments, the computational apparatus employed to perform the analysis is leased or rented. The computational resources may be part of an internet accessible collection of processors such as processing resources colloquially known as the cloud.
In some cases, the computations are performed by a parallel or massively parallel group of processors that are affiliated or liated with one another. The processing may be lished using distributed processing such as cluster computing, grid computing, and the like. In such ments, a cluster or grid of computational resources collective form a super l computer composed of multiple processors or computers acting together to perform the analysis and/or derivation described herein. These technologies as well as more conventional supercomputers may be employed to process sequence data as described herein. Each is a form of parallel ing that relies on processors or computers. In the case of grid computing these processors (often whole computers) are connected by a network (private, public, or the Internet) by a conventional network protocol such as Ethernet.
By contrast, a supercomputer has many processors connected by a local high-speed computer bus.
] In certain embodiments, the diagnosis (e.g., the fetus has Downs syndrome or the patient has a ular type of cancer) is generated at the same location as the analyzing ion. In other embodiments, it is performed at a different location. In some examples, reporting the diagnosis is performed at the location where the sample was taken, although this need not be the case. Examples of locations where the diagnosis can be generated or reported and/or where developing a plan is performed include health practitioners' offices, clinics, internet sites accessible by computers, and ld devices such as cell phones, tablets, smart phones, etc. having a wired or ss connection to a k. Examples of locations where counseling is performed include health practitioners' offices, clinics, internet sites accessible by computers, handheld devices, etc.
] In some embodiments, the sample collection, sample processing, and sequencing operations are performed at a first location and the analyzing and deriving operation is med at a second location. However, in some cases, the sample collection is collected at one location (e.g., a health practitioner's office or clinic) and the sample sing and cing is performed at a different on that is optionally the same on where the analyzing and ng take place.
In various embodiments, a sequence ofthe above-listed operations may be triggered by a user or entity initiating sample collection, sample processing and/or sequencing. After one or more these operations have begun execution the other ions may naturally follow. For example, the sequencing operation may cause reads to be automatically collected and sent to a processing apparatus which then ts, often automatically and possibly without further user intervention, the sequence analysis and derivation of aneuploidy operation. In some implementations, the result of this processing operation is then automatically delivered, possibly with reformatting as a diagnosis, to a system component or entity that processes reports the information to a health professional and/or patient. As explained such information can also be automatically processed to produce a treatment, testing, and/or monitoring plan, possibly along with counseling information. Thus, initiating an early stage operation can trigger an end to end sequence in which the health professional, patient or other concerned party is provided with a diagnosis, a plan, counseling and/or other information useful for acting on a physical condition. This is accomplished even though parts of the overall system are physically separated and possibly remote from the location of, e.g., the sample and ce apparatus.
Figure 5 shows one implementation of a dispersed system for producing a call or sis from a test sample. A sample collection location 01 is used for obtaining a test sample from a patient such as a nt female or a putative cancer patient. The samples then provided to a processing and cing location 03 where the test sample may be processed and sequenced as described above. Location 03 includes apparatus for processing the sample as well as apparatus for sequencing the processed sample. The result of the sequencing, as described elsewhere herein, is a collection of reads which are typically provided in an electronic format and ed to a network such as the Internet, which is indicated by reference number 05 in Figure 5.
WO 36059 The sequence data is provided to a remote location 07 where analysis and call generation are performed. This on may include one or more powerful computational devices such as computers or processors. After the computational resources at location 07 have completed their analysis and ted a call from the sequence information received, the call is d back to the network 05. In some implementations, not only is a call generated at location 07 but an associated diagnosis is also generated. The call and or sis are then transmitted across the network and back to the sample collection location 01 as rated in Figure 5. As explained, this is simply one of many variations on how the various operations associated with generating a call or diagnosis may be divided among various locations. One common t involves providing sample collection and processing and sequencing in a single location. r variation involves providing processing and sequencing at the same location as analysis and call generation.
Figure 6 elaborates on the options for performing various operations at distinct locations. In the most granular sense depicted in Figure 6, each of the following ions is performed at a separate location: sample collection, sample processing, sequencing, read alignment, calling, diagnosis, and reporting and/or plan development.
In one embodiment that aggregates some of these operations, sample processing and cing are performed in one location and read alignment, calling, and diagnosis are performed at a separate location. See the portion of Figure 6 identified by reference character A In another implementation, which is identified by character B in Figure 6, sample tion, sample processing, and sequencing are all performed at the same location. In this implementation, read alignment and calling are performed in a second on. y, diagnosis and reporting and/or plan development are performed in a third location. In the implementation depicted by character C in Figure 6, sample collection is performed at a first location, sample processing, sequencing, read alignment, calling,, and diagnosis are all performed together at a second location, and reporting and/or plan pment are performed at a third location. y, in the implementation labeled D in Figure 6, sample collection is performed at a first location, sample processing, sequencing, read alignment, and calling are all performed at a second location, and diagnosis and reporting and/or plan management are performed at a third location.
One embodiment provides a system for use in determining the presence or absence of any one or more different complete fetal chromosomal aneuploidies in a al test sample comprising fetal and maternal nucleic acids, the system including a sequencer for receiving a nucleic acid sample and providing fetal and maternal nucleic acid sequence information from the ; a sor; and a machine readable storage medium comprising instructions for ion on said processor, the instructions comprising: (a) code for obtaining sequence information for said fetal and maternal nucleic acids in the ; (b) code for using said sequence information to computationally fy a number of sequence tags from the fetal and maternal nucleic acids for each of any one or more chromosomes of st selected from chromosomes 1-22, X, and Y and to identify a number of sequence tags for at least one normalizing chromosome sequence or normalizing chromosome segment sequence for each of said any one or more somes ofinterest; (c) code for using said number of sequence tags identified for each of said any one or more chromosomes of interest and said number of sequence tags identified for each normalizing chromosome sequence or normalizing chromosome segment sequence to calculate a single chromosome dose for each of the any one or more chromosomes of interest; and (d) code for comparing each of the single chromosome doses for each of the any one or more chromosomes of interest to a corresponding old value for each of the one or more chromosomes of interest, and thereby determining the presence or absence of any one or more complete different fetal chromosomal aneuploidies in the sample.
In some embodiments, the code for calculating a single chromosome dose for each of the any one or more chromosomes of interest comprises code for ating a chromosome dose for a selected one of the chromosomes of interest as the ratio of the number of sequence tags identified for the selected chromosome of interest and the number of sequence tags identified for a corresponding at least one normalizing chromosome sequence or normalizing chromosome t ce for the selected chromosome of interest.
In some embodiments, the system further comprises code for repeating the calculating of a chromosome dose for each of any remaining chromosome segments of the any one or more segments of any one or more chromosomes of interest.
[00511] In some embodiments, the one or more chromosomes of interest selected from chromosomes 1-22, X, and Y comprise at least twenty chromosomes selected from somes 1-22, X, and Y, and wherein the instructions comprise instructions for ining the presence or absence of at least twenty different complete fetal chromosomal aneuploidies is determined.
[00512] In some embodiments, the at least one normalizing chromosome sequence is a group of somes selected from chromosomes 1-22, X, and Y. In other ments, the at least one normalizing chromosome sequence is a single chromosome selected from chromosomes 1-22, X, and Y.
Another ment provides a system for use m determining the presence or absence of any one or more different partial fetal chromosomal aneuploidies in a al test sample comprising fetal and maternal nucleic acids, the system comprising: a sequencer for receiving a nucleic acid sample and providing fetal and maternal c acid sequence ation from the sample; a sor; and a machine readable storage medium comprising instructions for execution on said processor, the instructions comprising: (a) code for obtaining sequence information for said fetal and maternal nucleic acids in said sample; (b) code for using said sequence information to computationally identify a number of sequence tags from the fetal and maternal nucleic acids for each of any one or more segments of any one or more chromosomes of interest selected from somes 1- 22, X, and Y and to identify a number of sequence tags for at least one izing segment sequence for each of said any one or more segments of any one or more chromosomes ofinterest; (c) code using said number of sequence tags identified for each of said any one or more segments of any one or more chromosomes of interest and said number of sequence tags identified for said normalizing segment sequence to calculate a single chromosome segment dose for each of said any one or more segments of any one or more chromosomes ofinterest; and (d) code for comparing each of said single chromosome segment doses for each of said any one or more segments of any one or more chromosomes of interest to a corresponding threshold value for each of said any one or more chromosome segments of any one or more some of interest, and thereby determining the presence or absence of one or more different partial fetal chromosomal aneuploidies in said sample.
In some embodiments, the code for calculating a single some segment dose comprises code for calculating a chromosome segment dose for a selected one of the chromosome segments as the ratio of the number of sequence tags identified for the ed chromosome segment and the number of sequence tags identified for a corresponding izing t sequence for the selected chromosome segment.
[00515] In some embodiments, the system further comprises code for repeating the calculating of a chromosome segment dose for each of any ing chromosome segments of the any one or more segments of any one or more somes rest.
In some embodiments, the system further es (i) code for repeating (a)-(d) for test samples from different maternal subjects, and (ii) code for determining the presence or absence of any one or more different partial fetal chromosomal oidies in each of said samples.
In other embodiments of any of the systems provided herein, the code further comprises code for automatically recording the ce or absence of a fetal chromosomal aneuploidy as determined in (d) in a patient l record for a human subject providing the maternal test sample, wherein the recording is performed using the processor.
In some embodiments of any of the systems provided herein, the sequencer 1s configured to perform next generation sequencing (NGS). In some embodiments, the sequencer is configured to perform massively parallel cing usmg sequencing-by-synthesis with reversible dye terminators. In other embodiments, the sequencer is configured to perform sequencing-by-ligation. In yet WO 36059 other embodiments, the sequencer 1s configured to perform single molecule sequencmg.
EXPERIMENTAL Example 1 Preparation and sequencing of primary and ed sequencing libraries a. Preparation ofsequencing libraries - iated protocol (ABB) All sequencing ies, i.e., pnmary and enriched libraries, were ed from approximately 2 ng of purified cIDNA that was extracted from maternal plasma. Library preparation was performed using reagents of the NEBNext™ DNA Sample Prep DNA Reagent Set 1 (Part No. E6000L; New England Biolabs, Ipswich, MA), for Illumina® as follows. Because cell-free plasma DNA is fragmented in nature, no further fragmentation by nebulization or sonication was done on the plasma DNA samples. The overhangs of approximately 2 ng purified cIDNA fragments contained in 40 µl were converted into phosphorylated blunt ends according to the t® End Repair Module by incubating in a 1.5 ml microfuge tube the cIDNA with 5µl 1OX phosphorylation buffer, 2 µl deoxynucleotide solution mix (10 mM each dNTP), lµl of a 1:5 dilution of DNA Polymerase I, 1 µl T4 DNA Polymerase and 1 µl T4 Polynucleotide Kinase provided in the NEBNext™ DNA Sample Prep DNA Reagent Set 1 for 15 minutes at 20°C. The enzymes were then heat inactivated by incubating the reaction mixture at 75°C for 5 minutes. The mixture was cooled to 4°C, and dA g ofthe blunt-ended DNA was accomplished using 10µ1 ofthe dA-tailing master mix containing the Klenow fragment (3' to 5' exo minus) (NEBNext™ DNA Sample Prep DNA Reagent Set 1), and incubating for 15 s at 37°C. Subsequently, the Klenow fragment was heat inactivated by incubating the reaction mixture at 75°C for 5 minutes. ing the inactivation of the Klenow fragment, 1 µl of a 1:5 dilution of Illumina Genomic Adaptor Oligo Mix (Part No. 1000521; Illumina Inc., d, CA) was used to ligate the Illumina adaptors ndex tors) to the dA-tailed DNA using 4 µl of the T4 DNA ligase provided in the NEBNext™ DNA Sample Prep DNA Reagent Set 1, by incubating the reaction mixture for 15 minutes at 25°C. The mixture was cooled to 4°C, and the adaptor-ligated cIDNA was purified from unligated adaptors, adaptor dimers, and other ts using magnetic beads provided in the Agencourt AMPure XP PCR purification system (Part No. A6388 l; Beckman Coulter Genomics, Danvers, MA). Eighteen cycles ofPCR were performed to selectively enrich adaptorligated cIDNA (25 µl) using Phusion ®High-Fidelity Master Mix (25µ1; Finnzymes, Woburn, MA) and na's PCR primers (0.5 µM each) complementary to the adaptors (Part No. 1000537 and 1000537). The adaptor-ligated DNA was subjected to PCR (98°C for 30 s; 18 cycles of 98°C for 10 seconds, 65°C for 30 seconds, and 72°C for 30; final ion at 72°C for 5 minutes, and hold at 4°C) using Illumina Genomic PCR s (Part Nos. 100537 and 1000538) and the Phusion HF PCR Master Mix provided in the t™ DNA Sample Prep DNA t Set 1, according to the manufacturer's instructions. The amplified product was purified using the Agencourt AMPure XP PCR purification system (Agencourt ence Corporation, Beverly, MA) according to the manufacturer's instructions available at www.beckmangenomics.com/products/AMPureXPProtocol_000387v001.pdf. The ed amplified product was eluted in 40 µl of Qiagen EB Buffer, and the concentration and size distribution of the amplified libraries was analyzed using the Agilent DNA 1000 Kit for the 2100 Bioanalyzer (Agilent technologies Inc., Santa Clara, CA). b. Preparation ofsequencing libraries -full-length protocol
[00520] The full-length ol described here is essentially the standard protocol provided by Illumina, and only differs from the na protocol in the cation of the amplified library. The Illumina protocol instructs that the amplified library be purified using gel electrophoresis, while the protocol described herein uses magnetic beads for the same purification step. Approximately 2 ng of purified cIDNA extracted from maternal plasma was used to prepare a primary sequencing library using NEBNext™ DNA Sample Prep DNA Reagent Set 1 (Part No. E6000L; New England Biolabs, Ipswich, MA) for Illumina® essentially according to the cturer's instructions. All steps except for the final purification of the adaptor-ligated products, which was performed using Agencourt magnetic beads and reagents instead of the purification column, were performed according to the protocol accompanying the NEBNext™ Reagents for Sample Preparation for a genomic DNA library that is sequenced using the Illumina® GAii.
The NEBNext™ ol essentially follows that provided by Illumina, which is available at grcf.jhml.edu/hts/protocols/11257047_ChIP_Sample_Prep.pdf.
The overhangs of approximately 2 ng purified cIDNA fragments contained in 40 µl were converted into phosphorylated blunt ends according to the NEBNext® End Repair Module by incubating the 40µ1 cIDNA with 5µ1 lOX phosphorylation buffer, 2 µl deoxynucleotide solution mix (10 mM each dNTP), 1 µl of a 1:5 dilution of DNA rase I, 1 µl T4 DNA Polymerase and 1 µl T4 cleotide Kinase provided in the NEBNext™ DNA Sample Prep DNA Reagent Set 1 in a 200 µl microfuge tube in a thermal cycler for 30 minutes at 20°C. The sample was cooled to 4°C, and purified using a QIAQuick column provided in the QIAQuick PCR Purification Kit (QIAGEN Inc., ia, CA) as follows. The 50 µl reaction was erred to 1.5 ml microfuge tube, and 250 µl of Qiagen Buffer PB were added. The resulting 300 µl were transferred to a QIAquick column, which was centrifuged at 13,000 RPM for 1 minute in a microfuge. The column was washed with 750 µl Qiagen Buffer PE, and re-centrifuged. Residual ethanol was removed by an onal centrifugation for 5 minutes at 13,000 RPM. The DNA was eluted in 39 µl Qiagen Buffer EB by fugation. dA tailing of 34 µl of the blunt-ended DNA was accomplished using 16 µl of the dA-tailing master mix ning the Klenow fragment (3' to 5' exo minus) (NEBNext™ DNA Sample Prep DNA Reagent Set 1), and incubating for 30 minutes at 37°C according to the manufacturer's t® dA-Tailing Module. The sample was cooled to 4°C, and purified using a column provided in the MinElute PCR Purification Kit (QIAGEN Inc., ia, CA) as follows. The 50 µl reaction was erred to 1.5 ml microfuge tube, and 250 µl of Qiagen Buffer PB were added. The 300 µl were transferred to the MinElute column, which was centrifuged at 13,000RPM for 1 minute in a microfuge. The column was washed with 750 µl Qiagen Buffer PE, and re-centrifuged. Residual ethanol was removed by an additional centrifugation for 5 minutes at 13,000 RPM. The DNA was eluted in 15 µl Qiagen Buffer EB by centrifugation. Ten microliters of the DNA eluate were incubated with 1 µl of a 1:5 dilution of the Illumina Genomic Adapter Oligo Mix (Part No. 1000521), 15 µl of 2X Quick on Reaction Buffer, and 4 µl Quick T4 DNA Ligase, for 15 minutes at 25°C according to the NEBNext® Quick Ligation Module. The sample was cooled to 4°C, and purified using a MinElute column as follows. One hundred and fifty microliters of Qiagen Buffer PE were added to the 30 µl reaction, and the entire volume was transferred to a MinElute column were transferred to a MinElute column, which was centrifuged at 13,000RPM for 1 minute in a microfuge. The column was washed with 750 µl Qiagen Buffer PE, and re-centrifuged. Residual ethanol was d by an additional centrifugation for 5 s at 13,000 RPM. The DNA was eluted in 28 µl Qiagen Buffer EB by centrifugation. Twenty three microliters of the adaptor-ligated DNA eluate were subjected to 18 cycles of PCR (98°C for 30 seconds; 18 cycles of 98°C for 10 seconds, 65°C for 30 seconds, and 72°C for 30; final ion at 72°C for 5 minutes, and hold at 4°C) using Illumina Genomic PCR s (Part Nos. 100537 and 1000538) and the n HF PCR Master Mix provided in the NEBNext™ DNA Sample Prep DNA Reagent Set 1, according to the manufacturer's instructions. The amplified product was purified using the Agencourt AMPure XP PCR purification system (Agencourt Bioscience Corporation, Beverly, MA) according to the manufacturer's instructions available at www.beckmangenomics.com/products/AMPureXPProtocol_000387v001.pdf. The Agencourt AMPure XP PCR purification system removes unincorporated dNTPs, primers, primer dimers, salts and other contaminates, and recovers amplicons greater than 100 bp. The purified amplified product was eluted from the Agencourt beads in 40 µl of Qiagen EB Buffer and the size distribution of the libraries was analyzed using the Agilent DNA 1000 Kit for the 2100 Bioanalyzer (Agilent technologies Inc., Santa Clara, CA). c. Analysis ofsequencing ies prepared according to the abbreviated (a) and the full-length (b) protocols The electropherograms generated by the Bioanalyzer are shown in Figures 7A and 7B. Figure 7A shows the electropherogram of library DNA ed from cIDNA purified from plasma sample M24228 using the full-length protocol described in (a), and Figure 7B shows the electropherogram of library DNA ed from cIDNA purified from plasma sample M24228 using the full-length protocol bed in (b ). In both figures, peaks 1 and 4 represent the 15 bp Lower Marker, and the 1,500 Upper Marker, respectively; the numbers above the peaks indicate the migration times for the library fragments; and the horizontal lines indicate the set old for integration. The electropherogram in Figure 7A shows a minor peak of fragments of 187 bp and a major peak of fragments of 263 bp, while the opherogram in Figure 7B shows only one peak at 265 bp. Integration of the peak areas resulted in a calculated concentration of 0.40 ng/µl for the DNA ofthe 187 bp peak in Figure 7A, a concentration of 7.34 ng/µl for the DNA ofthe 263bp peak in Figure 7A, and a concentration of 14.72 ng/µl for the DNA of the 265 bp peak in Figure 7B. The Illumina adaptors that were ligated to the cIDNA are known to be 92 bp, which when subtracted from the 265 bp, indicate that the peak size of the cIDNA is 173 bp. It is possible that the minor peak at 187 bp represents nts of two primers that were ligated end-to-end. The linear two-primer fragments are ated from the final library product when the abbreviated protocol is used. The abbreviated protocol also eliminates other smaller fragments of less than 187 bp. In this example, the concentration of purified adaptor-ligated cIDNA is double that of the adaptorligated cIDNA produced using the full-length protocol. It has been noted that the concentration of the adaptor-ligated cIDNA fragments was always r than that obtained using the full-length protocol (data not shown).
] Thus, an advantage of preparing the sequencing library usmg the abbreviated protocol is that the library obtained consistently comprises only one major peak in the 262-267 bp range while the quality ofthe library prepared using the full-length ol varies as reflected by the number and mobility s other than that representing the cIDNA. Non-cIDNA products would occupy space on the flow cell and diminish the quality of the r amplification and subsequent imaging of the sequencing reactions, which underlies the overall assignment of the aneuploidy status. The abbreviated protocol was shown not to affect the sequencing of the library.
Another advantage of prepanng the sequencmg library usmg the abbreviated protocol is that the three enzymatic steps t-ending, d-A tailing, and adaptor-ligation, take less than an hour to te to support the validation and implementation of a rapid aneuploid diagnostic service. ] r advantage is that the three enzymatic steps of blunt-ending, d- A tailing, and adaptor ligation, are performed in the same reaction tube, thus avoiding multiple sample transfers that would potentially lead to loss of material, and more importantly to possible sample mix-up and sample contamination.
WO 36059 Example 2 Non-Invasive Prenatal Testing Using Fragment Size Introduction ] Since its commercial introduction in late 2011 and early 2012, non- invasive prenatal testing (NIPT) of cell free DNA (cIDNA) in maternal plasma has rapidly become the method of choice to screen nt women at high risk for fetal aneuploidies. The methods are based primarily on ing and sequencing cIDNA in the plasma of pregnant women, and ng the number of cIDNA fragments that align to particular regions of the reference human genome (references: Fan et al., Lo et al.). These DNA sequencing and molecular counting methods allow a high precision determination of the relative copy numbers for each of the chromosomes across the genome. High sensitivities and specificities for the detection of trisomies 21, 18 and 13 have been reproducibly achieved in multiple clinical studies (refs, cite Gil/Nicolaides meta-analysis).
[00527] More recently, additional clinical studies have shown that this approach can be extended to the l obstetric population. There are no able differences in the fetal fractions between the high- and average-risk populations (refs).
Clinical study results demonstrate that NIPT using molecular counting by cIDNA sequencing performs equivalently in both populations. A statistically significant improvement in the positive predictive value (PPV) over standard serum screening has been trated (refs). Lower false positive test results, as compared with serum mistry and nuchal translucency measurement, have significantly reduced the need for invasive diagnostic procedures (see Larion et al. references from ad's group).
[00528] Given the good NIPT performance in the general obstetric population, workflow simplicity and costs have now become a main consideration for the implementation of cIDNA sequencing for whole chromosome aneuploidy detection in the general obstetric population ence: ISPD Debate 1, Brisbane). Most NIPT tory s utilize a polymerase chain reaction (PCR) amplification step after the library preparation and single end sequencing that requires 10-20 million unique cIDNA fragments to achieve reasonable sensitivity to detect aneuploidy. The complexity of the PCR based ow and deeper sequencing requirements have limited the potential ofthe NIPT assay and have resulted in increased costs.
Here it is demonstrated that high analytical sensitivities and specificities can be achieved with a simple library preparation using very low cIDNA input that does not require PCR amplification. The PCR free method simplifies the workflow, improves the turnaround time and eliminates biases that are inherent with PCR methods. The amplification free workflow can be coupled with paired end cing to allow determination of fragment length for each tag and the total fetal fraction in each . Since the fetal cIDNA fragments are shorter than maternal fragments [ref Quake 2010, should also cite Lo's Science Clin Translation article], the detection of fetal aneuploidy from maternal plasma can be made much more robust and efficient, requiring fewer unique cIDNA nts. In combination, improved analytical sensitivity and specificity is achieved with a very fast turnaround time at a significantly lower number of cIDNA fragments. This potentially allows NIPT to be carried out at significantly lower costs to facilitate application in the l ric population.
Methods Peripheral blood samples were drawn into BCT tubes (Streck, Omaha, NE, USA) and shipped to the Illumina CLIA laboratory in Redwood City for commercial NIPT testing. Signed patient consent forms permitted second plasma aliquots to be de-identified and utilized for clinical research, with the exception of patient samples sent from the state ofNew York. Plasma s for this work were selected to include both unaffected and aneuploid fetuses with a range of cIDNA concentrations and fetal fractions.
Simplification ofLibrary Processing cIDNA was ted from 900µL of maternal plasma usmg the NucleoSpin l blood purification kit (Macherey-Nagel, Duren, Germany) with minor modifications to accommodate a larger lysate input. The isolated cIDNA was put directly into the sequencing y process without any normalization of the cIDNA input. Sequencing libraries were prepared with a TruSeq PCR Free DNA library kit (Illumina, San Diego, CA, USA) with dual indexes for barcoding the 2016/067886 cIDNA fragments for sample identification. The following modifications to the library protocol were used to improve the compatibility ofthe library preparation with the low concentration of input cIDNA. Template input volume was increased, while the end repair, A-tailing and ligation master mix and adapter concentrations were decreased. Additionally, after end repair, a heat kill step was introduced to deactivate enzymes, the post end repair SPRI (vendor) bead purification step was d, and elution during the post ligation SPRI bead purification step utilized HTl buffer (Illumina).
A single MICROLAB® STAR (Hamilton, Reno, NV, USA) liquid handler, ured with a 96 channel head and 8 1-mL pipetting channels, was used to batch process 96 plasma samples at a time. The liquid handler processed each individual plasma sample through DNA extraction, sequencing library preparation and quantitation. Individual sample libraries were quantified with AccuClear (Biotium, Hayward, CA, USA) and pools of 48 samples were prepared with normalized inputs ing in a final tration of 32 pM for sequencing.
PairedEndSequencing ] DNA sequencing was carried out with an Illumina NextSeq 500 instrument utilizing 2x36 bp paired end sequencing, plus an additional 16 cycles for sequencing the sample es. A total of 364 samples were run across 8 independent sequencing batches.
Paired DNA sequences were de-multiplexed using bcl2fastq (Illumina) and mapped to the reference human genome (hgl9) using bowtie2 aligner algorithm [ref ad]. Paired reads had to match forward and reverse strands to be d. All counted mapped pairs exceeding mapping quality scores of 10 (Ruan et al.) with globally unique first reads were ed to non-overlapping consecutive fixed-width genomic bins of 100 kb in size. Approximately 2% of the genome showed highly variable coverage across an independent set of NIPT s and was excluded from further analysis.
Using genomic location information and nt size available from mapped locations of each of the two ends of the sequenced cIDNA fragments, two variables were d for each 100 kb window: (a) total counts of short fragments below 150 base pairs in length, and (b) fraction of fragments between 80 and 150 base pairs within the set of all fragments below 250 base pairs. Limiting the size of fragments to less than 150 base pairs enriches for fragments originating from the placenta, which is a proxy for fetal DNA The fraction of short fragments characterizes the relative fetal cIDNA amounts in the plasma mixture. CIDNA from a trisomic fetus would be ed to have a higher on of short reads mapping to the trisomic chromosome compared to a euploid fetus that is disomic for that chromosome.
The counts and fractions of short nts were independently normalized to remove systematic assay biases and sample-specific ions attributable to genomic guanine cytosine (GC) content utilizing the s shown in Figure 2D. Normalized values were trimmed by removing bins deviating from the whole chromosome median by more than 3 robust es of standard deviation.
Finally, for each of the two variables, trimmed normalized values associated with the target chromosome were compared to those on normalizing reference somes to construct at-statistic.
Data from each paired end cing run followed four steps for analysis: 1) read conversion, 2) feature binning at lOOkb resolution, 3) normalization of each e (counts and on) at lOOkb resolution and 4) combining features and scoring for aneuploidy detection. In step 1, sample data is de-multiplexed from the individual barcodes, aligned to the genome and filtered for sequence quality. Step 2 total counts of short fragments below 150 base pairs in length, and fraction of fragments between 80 and 150 base pairs within the set of all fragments below 250 base pairs are determined for each bin. Assay bias and sample specific variations are removed in step 3. Finally, enrichment over a reference is determined and scored using a t-test for each of the counts and fraction, and combined for final score for aneuploidy ion.
Detection o[Fetal Whole Chromosome Aneuploidv We tested whether the counts and fraction data could be combined to enhance the ability to detect fetal trisomy 21. Sixteen plasma samples from pregnant women ng fetuses with karyotype-confirmed trisomy 21 and 294 samples from unaffected pregnancies were randomly distributed across processing batches, resulting in nine flow cells for sequencing. Each algorithm step was examined separately to WO 36059 determine the ability of each step and combination of steps to detect aneuploidy. The final score for fetal oidy detection in the combined case was defined as the square root of the sum of squares of the two individual t-statistics, and a single threshold was applied to generate a call of "aneuploidy detected" versus "aneuploidy not detected".
Calculation ofFetal Fraction For each sample, fetal fraction was ted using a ratio of the total number of fragments of size [111, 13 6 bp] to the total number of fragments of size [165,175 bp] within a subset ofthe genomic 100 kb bins. Using samples from women ng known male fetuses, the top 10% of genomic bins that had the highest correlation with fetal fraction derived from the number of copies of the X some [ref Rava] were determined. The correlation between fragment sizebased fetal fraction estimates and those derived from the X chromosome in known male fetuses was computed using a leave-one-out cross validation [REF] analysis that included both bin selection and regression model parameter estimation. The estimated fetal fraction was then derived using a linear regression model from the fragment size ratios.
Results Simplification ofLibrary sing
[00540] Figure 8 shows the overall workflow and timeline for this new version of NIPT compared to the standard tory workflow. The entire 96-sample ation workflow for plasma isolation, cIDNA tion, library construction, quantitation, and pooling was able to process samples in less than 6 hours total preparation time on a single Hamilton STAR. This compares to 9 hours and two Hamilton STARs with the PCR based methods used in the CLIA laboratory. The amount of cIDNA extracted per sample averaged 60 pg/µL, and the yield of the sequencing library output was linearly ated (R2=0.94) with cIDNA input as shown in Figure 9. The average recovery was greater than 70% (add range), indicating a highly efficient recovery of the cIDNA after the SPRI bead purification.
Each sequencing run used normalized amounts of 48 s multiplexed and took approximately 14 hours to complete. The median number of uniquely mapped paired reads was XXX M with 95% of s above YYY.
PairedEndSequencing The total sequencing time per 48-sample batch was less than 14 hours on the NextSeq 500. This compares to a 40 hours (1 flow cell, 96 samples) or 50 hours (2 flow cells, 192 samples) for the laboratory process on a HiSeq 2500. The mapped genomic locations of both ends of cIDNA fragments provided cIDNA fragment size information. Figure 10 shows the cIDNA nt size distribution as measured from 324 samples from pregnancies with a male fetus. The size of nts that mapped to autosomal chromosomes known to be d and primarily represent the maternal chromosomes is represented by the thin curve. The average size of the inserts was 175 bp with XX% of nts measuring between 1OObp and 200bp. The thick curve represents the fragment size that exclusively arises from the Y-chromosome representing only fetal cIDNA nts. The size distribution from the Y-chromosome specific sequences was smaller, averaging 167 bp with a 10 base periodicity at r fragment sizes.
Since the shorter fragments of cIDNA are enriched for fetal DNA, selective analysis using only shorter fragments would be expected to increase the relative fetal representation due to preferential selection of fetal reads. Figure 11 shows the relative fetal fraction from the total counts of mapped paired end reads ed to the counts from paired end reads that are less than 150 bp. Overall, the median fetal fraction increases by a factor of 2 ed to the total counts albeit with some increase in the variance. The size cutoff of 150 bp was found to provide an optimum tradeoff for counts with an increase in fetal representation versus variance in the counts.
Detection o[Fetal Whole Chromosome Aneuploidv Each of the available metrics, total counts, counts less than 150 bp, fraction of counts enriched for fetal cIDNA (counts between 80 and 150 nts <250bp) and the combination ofthe shorter fragment counts with fraction, were tested for the ability to differentiate trisomy 21 samples from those euploid in chromosome 21. Figure 12 shows the results for each of these metrics. The total counts have a median of XX counts while the counts less than 150 bp has a median of YY counts.
Yet, as can be seen in Fig 4A and 4B, the smaller counts show better separation between trisomy 21 and d ily because this metric is enriched for the fetal cIDNA. The fraction alone is nearly as effective as the total counts for differentiating aneuploidy (Fig. 4C), but when used in combination with the short fragment counts (Fig. 4D) provides improved differentiation over short fragment counts alone. This indicates that the fraction is providing independent information that enhances the detection of trisomy 21. When compared to the t CLIA laboratory workflow using library prep with PCR amplification and a median of 16 M counts/sample, the PCR free, paired end sequencing workflow shows equivalent performance with significantly fewer counts/sample (e.g., 6 M counts/sample or fewer) and a simpler, shorter sample preparation workflow.
Calculation ofFetal Fraction Using the X chromosome results from pregnancies with male fetuses, normalized chromosome values can be utilized to determine fetal fractions for the counts (ClinChem ref) and compared for different cIDNA fragment sizes. Fetal fractions derived from the X chromosome were used to calibrate the ratios for a set of 140 samples and estimate mance using a one-out cross-validation. Figure 13 shows the results of cross-validated fetal fraction predictions and demonstrates the correlation n the two data sets, indicating that fetal fraction estimates can be obtained from any samples, including ones from women carrying female fetuses once a calibration set has been measured.
Discussion It has been demonstrated that high ical sensitivity and icity for fetal aneuploidy detection from cIDNA in al plasma can be achieved with a PCR free library preparation coupled with paired end DNA sequencing. The method simplifies workflow, improves turnaround time e 8) and should eliminate some biases inherent with PCR methods. Paired end sequencing allows determination of fragment length sizes and fetal fraction that can be further utilized to enhance detection of aneuploidy at significantly lower tag counts ed to currently implemented commercial methods. Performance of the PCR free paired end implementation appears to be similar to single end sequencing methods that utilize up to three times the number oftags.
WO 36059 Simplification ofLibrary Processing The PCR free workflow has l advantages for the clinical laboratory. Because of the high yield and linear behavior of the library preparation, normalized pools of samples for sequencing can be made directly from the individual sample library concentrations. Biases inherent in the PCR amplification ofthe library preparation process are thereby eliminated. In addition, there is no need to isolate separate liquid rs for pre- and post-PCR ties; this reduces the capital burden for the laboratory. This simplified workflow allows batches of samples to be prepared within a single shift of the clinical laboratory, and then sequenced and analyzed overnight. Overall, the reduced capital expenditure, reduced "hands on" time and rapid turnaround allow for potentially significant reductions in the cost and overall robustness ofNIPT.
PairedEndSequencing Using paired end sequencing on the NextSeq 500 system has several advantages for the counting of cIDNA fragments. First, with dual index barcodes, samples can be multiplexed at a high level allowing normalization and correction of -run variation with high statistical confidence. In addition, because 48 samples are being multiplexed per run, and the amount needed on the flow cell for clustering is limited, the input requirement per sample is significantly reduced, ng the PCR free library ow to be utilized. With their typical cIDNA yield of approximately ng per sample, researchers were able to get 2-3 sequencing runs per sample even without PCR amplification. This is in contrast to other approaches that require significant amounts of plasma input from multiple blood tubes to yield enough cIDNA for aneuploidy determination (REF). Finally, paired end sequencing allows the determination of cIDNA fragment size and analytical enrichment for fetal cIDNA.
Detection o[Fetal Whole Chromosome Aneuploidv Our s demonstrate that counts of cIDNA fragments below 150 bp are able to better differentiate aneuploidy from euploid somes than the total counts. This observation is in contrast to the results of Fan et al., who suggested that the cy of the counting statistics would be decreased using r fragments (Fan et al.) because of the reduction the number of available counts. The fraction of short nts also provides some entiation for trisomy 21 detection as implied by Yu et al., albeit with less c range than the counts. However, combining the counting and fraction metrics results in the best tion of the trisomy 21 s from euploid, and implies that these two metrics are complementary measurements for chromosome entation. Other biological metrics, e.g. methylation, might also provide orthogonal information that could enhance the signal-to-noise ratio for oidy detection.
Calculation ofFetal Fraction The methods presented here also allow an estimation of the fetal on in each sample without creating additional laboratory work. With many samples on each flow cell, approximately half of which are male, an accurate fetal fraction estimate can be ed for all samples by calibrating fetal fraction measurement from fragment size information with that determined from the male samples. In the commercial setting, researchers' clinical experience has shown that standard ng methods using a larger number of single end tags has led to very low false negative rates even in the absence of specific fetal fraction measurements (REF). Given the similar limit of detection ed here, equivalent test performance is expected.
Conclusion It has been demonstrated that high analytical sensitivity and specificity for fetal aneuploidy detection from cIDNA in maternal plasma can be achieved with a PCR free y preparation coupled with paired end DNA sequencing. This simplified workflow has a very fast turnaround time, potentially allowing NIPT to be carried out at significantly lower cost for use in the general obstetric population. In addition, the paired end sequencing techniques have the ial to e other biological phenomena, as well as providing other clinical applications. For example, size information from methylated specific regions of the genome or CpG islands could provide another orthogonal metric for enhancing the detection of copy number variants across the genome.
The present disclosure may be embodied in other specific forms without ing from its spirit or essential teristics. The described embodiments are to be considered m all respects only as illustrative and not restrictive. The scope ofthe disclosure is, therefore, indicated by the appended claims rather than by the foregoing ption. All changes which come within the meaning and range of equivalency ofthe claims are to be embraced within their scope

Claims (21)

Claims
1. A method, implemented using a computer comprising one or more processors and system memory, for determining a copy number variation (CNV) of a nucleic acid sequence of interest in a test sample comprising cell-free nucleic acid fragments originating from two or more genomes, the method comprising: (a) determining fragment sizes of the cell-free nucleic acid fragments t in the test sample, wherein the cell-free nucleic acid fragments are not artificially fragmented; (b) determining first ges of sequence tags for the bins of a reference genome using ce tags for the cell-free c acid fragments having sizes in a first size domain, wherein the ce tags are obtained by aligning sequence reads of the cell-free nucleic acid fragments in the test sample to the reference genome; (c) determining second coverages of the sequence tags for the bins of the reference genome using sequence tags for the cell-free nucleic acid fragments having sizes in a second size domain, wherein the second size domain is ent from the first size domain; and (d) determining a copy number variation in the sequence of interest using a statistical value calculated from the first coverages and the second coverages.
2. A system for evaluation of copy number of a nucleic acid sequence of interest in a test sample comprising cell-free nucleic acid nts, the system sing: a processor; and one or more computer-readable storage media having stored thereon instructions for execution on said processor to: (a) determine fragment sizes of the cell-free nucleic acid fragments present in the test sample, wherein the cell-free nucleic acid nts are not artificially fragmented; (b) determine first ges of sequence tags for the bins of the reference genome using sequence tags for the cell-free nucleic acid fragments having sizes in a first size domain, wherein the sequence tags are obtained by aligning sequence reads of the cell-free nucleic acid fragments in the test sample to the reference genome; (c) ine second coverages of the sequence tags for the bins of the reference genome using ce tags for the cell-free nucleic acid fragments having sizes in a second size domain, wherein the second size domain is different from the first size domain; and (d) determine a copy number variation in the sequence of interest using a statistical value calculated from the first coverages and the second ges.
3. The system of claim 2, wherein a likelihood ratio is calculated from the first coverages and the second coverages for monosomy X, trisomy X, trisomy 13, trisomy 18, or trisomy 21.
4. The system of claim 2, further sing normalizing a number of the sequence tags by: normalizing for GC content of the sample, normalizing for a global wave profile of variation of a training set, and/or izing for one or more components obtained from a principal component analysis.
5. The system of claim 2, wherein a reference region is ed from the group consisting of: all robust chromosomes, robust chromosomes not including the sequence of interest, at least a chromosome e of the sequence of st, and a subset of chromosomes selected from the robust chromosomes.
6. The system of claim 5, wherein the reference region comprises robust chromosomes that have been determined to provide the best signal detection ability for a set of training samples.
7. The system of claim 2, further comprising: calculating values of a size parameter for the bins by, for each bin: (i) determining a value of the size parameter from sizes of cell-free nucleic acid fragments in the bin, and (ii) normalizing the value of the size parameter by accounting for bin-to-bin variations due to factors other than copy number variation; and determining a ased t-statistic for the sequence of interest using values of the size parameter of bins in the sequence of interest and values of the size parameter of bins in the reference region for the sequence of st.
8. The system of claim 3, further comprising obtaining a plurality of likelihood ratios and applying the plurality of hood ratios to a decision tree to determine a ploidy case for the sample.
9. The system of claim 2, wherein (d) comprises determining the copy number variation based on a likelihood ratio obtained from the first coverages and the second coverages.
10. The system of claim 9, wherein the likelihood ratio comprises a first likelihood that the test sample is an aneuploid sample over a second likelihood that the test sample is a euploid sample.
11. The system of claim 2, wherein (d) comprises determining the copy number variation in the sequence of interest using size characteristics of bins in the sequence of interest in addition to the first coverages and the second coverages, wherein the size characteristics of the bins were determined using nt sizes of reads aligned to the bins.
12. The system of claim 11, wherein the size characteristic for a bin comprises a ratio of nts of size r than a defined value to total fragments in the bin.
13. The system of claim 11, wherein (d) comprises calculating a likelihood ratio from a first tstatistic for the ce of interest using the first coverages, a second istic for the sequence of interest using the second coverages, and a third t-statistic for the sequence of interest using the size characteristics.
14. The system of claim 2, wherein the first size domain comprises ree nucleic acid fragments of substantially all sizes in the sample, and the second size domain comprises only cell-free nucleic acid fragments smaller than a defined size.
15. The system of claim 2, wherein the second size domain comprises only the cell-free nucleic acid nts smaller than about 150 bp.
16. The system of claim 2, wherein (f) comprises determining the copy number ion in the sequence of interest based on a likelihood ratio obtained using one or more values of fetal fraction , the first ges, and the second coverages.
17. The system of claim 16, wherein the one or more values of fetal fraction comprise a value of fetal fraction ated using the information about the sizes of the cell-free nucleic acid fragments.
18. The system of claim 17, wherein the value of fetal fraction is calculated by: obtaining a frequency distribution of the fragment sizes; and applying the frequency bution to a model relating fetal fraction to frequency of fragment size to obtain the fetal fraction value.
19. The system of claim 18, wherein the one or more values of fetal fraction comprise a value of fetal fraction calculated using coverage information for the bins of the nce .
20. The system of claim 19, wherein the value of fetal fraction is ated by: applying coverage values of a plurality of bins to a model relating fetal fraction to coverage of bin to obtain the fetal fraction value.
21. A computer program product comprising one or more computer-readable non-transitory storage media having stored thereon computer-executable instructions that, when executed by one or more processors of a computer system, cause the computer system to: (a) determine fragment sizes of the cell-free nucleic acid fragments present in the test sample, wherein the cell-free nucleic acid fragments are not artificially fragmented; (b) determine first coverages of sequence tags for the bins of a reference genome using sequence tags for the cell-free nucleic acid fragments having sizes in a first size domain, n the sequence tags are obtained by aligning sequence reads of the cell-free nucleic acid fragments in the test sample to the reference genome; (c) determine second coverages of the sequence tags for the bins of the reference genome using sequence tags for the ree nucleic acid fragments having sizes in a second size domain, wherein the second size domain is different from the first size domain; (d) determine a copy number ion in the sequence of interest using a statistical value calculated from the first coverages and the second coverages.
NZ752319A 2016-02-03 2016-12-20 Using cell-free dna fragment size to determine copy number variations NZ752319B2 (en)

Applications Claiming Priority (5)

Application Number Priority Date Filing Date Title
US201662290891P 2016-02-03 2016-02-03
US62/290,891 2016-02-03
US15/382,508 US10095831B2 (en) 2016-02-03 2016-12-16 Using cell-free DNA fragment size to determine copy number variations
US15/382,508 2016-12-16
NZ745637A NZ745637A (en) 2016-02-03 2016-12-20 Using cell-free dna fragment size to determine copy number variations

Publications (2)

Publication Number Publication Date
NZ752319A NZ752319A (en) 2021-01-29
NZ752319B2 true NZ752319B2 (en) 2021-04-30

Family

ID=

Similar Documents

Publication Publication Date Title
AU2019203491B2 (en) Using cell-free DNA fragment size to determine copy number variations
US20210371907A1 (en) Using cell-free dna fragment size to determine copy number variations
US11875899B2 (en) Analyzing copy number variation in the detection of cancer
US11697846B2 (en) Detecting and classifying copy number variation
US20240203601A1 (en) Analyzing copy number variation in the detection of cancer
NZ752319B2 (en) Using cell-free dna fragment size to determine copy number variations
EA045158B1 (en) USING CELL-FREE DNA FRAGMENT SIZE TO DETERMINE COPY NUMBER VARIATIONS