EA040939B1 - ANALYSIS OF THE FETUS GENOME FROM A MATERNAL BIOLOGICAL SAMPLE - Google Patents

Description

Cross-references to related applications

This application claims priority from and is a non-provisional application for U.S. Patent No. 61/258567, entitled Fetal Genome Analysis and filed Nov. 5, 2009, U.S. Provisional Application No. 61/259075 entitled Analysis of the Fetal Genome from a Maternal Biological Sample and filed Nov. 6, 2009, and U.S. Provisional Application No. 61/381854 entitled Analysis of the Fetal Genome from Maternal Biological Sample and filed September 10, 2010, the entire contents of which are incorporated by reference to all cases.

The present application is also related to U.S. Patent Application No. 12/178181 entitled Diagnosis of Fetal Chromosomal Aneuploidy by Massive Parallel Genome Sequencing and filed November 5, 2009 (Attorney Docket No. 016285-005220 US); US Patent Application No. 12/614350, entitled Diagnosis of fetal chromosomal aneuploidy by enriched genome sequencing (Attorney Docket No. 016285-005221 US), and co-pending US application entitled Size Based Genome Analysis (Attorney Docket No. 016285-006610 US ), all contents of which are incorporated by reference to all occurrences.

State of the art

The present invention relates generally to the analysis of a fetal genome based on a maternal sample, in particular the determination of all or part of the fetal genome based on the analysis of genetic fragments in a maternal sample.

The discovery of cell-free fetal nucleic acids in maternal plasma in 1997 opened up new possibilities for non-invasive prenatal diagnosis (Lo YMD et al., 1997, 350: 485-487; and US Patent 6285540). This technology quickly translated into clinical applications with the detection of paternally derived genes or sequences, for example, to determine the sex of the fetus, determine the RhD status of the fetus, and determine whether it inherited a paternally transmitted mutation (Amicucci P et al., Clin Chem 2000, 46: 301-302; Saito H et al., Lancet 2000, 356: 1170; and Chiu RWK et al., Lancet 2002, 360: 998-1000). Recent advances in this field have made it possible to prenatally diagnose chromosomal aneuploidy in trisomy 21 type fetuses from maternal plasma nucleic acid analysis (Lo YMD et al., Nat Med 2007, 13: 218-223; Tong YK et al., Clin Chem 2006, 52:2194-2202; US Patent Publication 2006/0252071; Lo YMD et al., Proc Natl Acad Sci USA 2007, 104: 13116-13121; Chiu RWK et al., Proc Natl Acad Sci USA 2008, 105: 20458-20463 ; Fan HC et al., Proc Natl Acad Sci 2008, 105: 16266-16271; US Patent Publication 2007/0202525; and US Patent Publication 2009/0029377).

Another area of significant recent progress is the use of single molecule enumeration techniques such as digital PCR for non-invasive prenatal diagnosis of monogenic diseases in which both mother and father carry the same mutation. This is achieved by analysis of relative mutation dose (RMD) in maternal plasma (US Patent Application 2009/0087847; Lun FMF et al., Proc Natl Acad Sci USA 2008, 105: 19920-19925; and Chiu RWK et al., Trends Genet 2009, 25: 324-331).

However, such methods use existing knowledge about possible mutations to analyze specific parts of the genome, so they cannot identify latent or rare mutations or genetic diseases. Therefore, it is necessary to provide new methods, systems and devices that can identify all or part of the genome in the fetus using non-invasive methods.

The essence of the invention

In some embodiments of the present invention, methods, systems, and devices are provided for determining at least a portion of the genome of an unborn fetus in a pregnant mother. It is possible to make a genetic map of the entire genome or specific regions of the fetal genome before birth using samples containing fetal and maternal genetic material (for example, a blood sample from a pregnant mother). The genetic map may contain sequences that the fetus has inherited from both parents, or only sequences from one of the parents. Based on one or more of these genetic maps, it is possible to determine the risk of a fetus developing a genetic disease or predisposition to a genetic or other disease or genetic trait. Other uses of the embodiments are also described.

In one embodiment, DNA fragments from a maternal sample (comprising maternal and fetal DNA) can be analyzed to identify alleles at certain predetermined loci (landmarks). You can then analyze the total number of DNA fragments of the corresponding alleles at these loci to determine the relative number of haplotypes at these loci and thereby determine which haplotypes are inherited by the fetus from the maternal and/or paternal genome. By identifying fetal haplotypes, the fetal genotype can be determined from individual loci in the appropriate genomic region, including target loci. In various embodiments, loci for which parents are represented by certain combinations of homozygous and heterozygous can be analyzed in such a way as to determine regions of the fetal genome. In one embodiment, control haplotypes representing common haplotypes in a population are used to determine maternal and paternal genomes, along with analysis of DNA fragments from a maternal sample. Other embodiments are also envisaged, such as the determination of mutations, the determination of the proportion of the fetus in the sample

- 1 040939 ri and determination of maternal sample sequencing coverage.

Other embodiments of the invention relate to systems, devices and computer-readable media associated with the methods described here. In one embodiment, the computer-readable medium contains instructions for acquiring data and analyzing data, but not instructions for directing a machine to generate data (eg, sequencing nucleic acid molecules). And in another embodiment, the computer-readable medium contains instructions for directing a machine to create data. In one embodiment, the computer program product includes a computer-readable medium storing a plurality of instructions for controlling a processor in performing operations for the methods described herein. Embodiments also relate to computer systems configured to perform the steps of any of the methods described herein, optionally with different components performing the respective step or respective group of steps.

Referring to the rest of the description, including the drawings and the claims, other features and advantages of embodiments of the present invention will be realized. Additional features and advantages, as well as the structure and operation of various embodiments of the present invention are described in detail below with the help of the accompanying drawings. In the figures, the same designation numbers may mean identical or functionally similar elements.

Brief description of the figures

In FIG. 1 is a flow diagram of a method for determining 100 at least a portion of the genome of an unborn fetus of a pregnant woman in accordance with embodiments of the present invention.

In FIG. 2 shows two paternal and two maternal haplotypes for a specific segment of their respective genomic code, according to embodiments of the present invention.

In FIG. 3 shows two types of SNPs in the parental haplotypes from FIG. 2 according to embodiments of the present invention.

In FIG. 4A and B are assays to determine fetal haplotypes for two types of SNPs according to embodiments of the present invention.

In FIG. 5A and B are comparisons of the relative abundance (eg, occurrence) of fragments at each locus and whether the result of the comparison classifies a particular genotype as heritable or not, according to embodiments of the present invention.

In FIG. 6 shows the effect of changing the probability ratio in SPRT classification according to embodiments of the present invention.

In FIG. 7 is a flow diagram of a method for determining 700 at least the paternally inherited portion of the genome in an unborn fetus of a pregnant woman, in accordance with embodiments of the present invention.

In FIG. 8 is a flow diagram of a method for determining 800 at least a portion of the genome in an unborn fetus of a pregnant woman at a site in which the father and mother are heterozygous, according to embodiments of the present invention.

In FIG. 9 shows haplotypes of a father and mother that are both heterozygous at a particular genomic site according to embodiments of the present invention.

In FIG. 10 is a flow diagram of a method for determining the 1000 fractional concentration of fetal material in a maternal sample, in accordance with embodiments of the present invention.

In FIG. 11 is a flow diagram of a method for determining that a locus is informative in accordance with embodiments of the present invention.

In FIG. 12A and B show the predicted distribution of occurrence for the T allele (the least common allele in scenarios (a) and (c)) over the three scenarios, assuming a fractional concentration of fetal DNA of 20% and 5%, respectively, according to embodiments of the present invention.

In FIG. 13A, B and 14 show the predicted distribution of occurrence for the least common allele at a fractional concentration of fetal DNA of 20% at various total number of molecules corresponding to SNPs, according to embodiments of the present invention.

In FIG. 15A and B are examples of control haplotypes, parental haplotypes derived from control haplotypes, and resulting fetal haplotypes according to embodiments of the present invention.

In FIG. 16 is a flow diagram of a method for determining 1600 at least a portion of the fetal genome when a set of control haplotypes is known but parental haplotypes are not known, in accordance with embodiments of the present invention.

In FIG. 17 shows an example of determining informative loci when analyzing DNA fragments from a maternal sample according to embodiments of the present invention.

In FIG. 18 shows the three control haplotypes (Hap A, Hap B and Hap C) and paternal alleles.

In FIG. 19 shows the determination of the parental haplotype from paternal alleles according to embodiments of the present invention.

In FIG. 20 shows the calculation of maternal genotypes from analysis of a maternal sample according to embodiments of the present invention.

- 2 040939

In FIG. 21 depicts an embodiment for determining maternal haplotypes from maternal genotypes and control haplotypes according to embodiments of the present invention.

In FIG. 22 shows maternal haplotypes and paternally inherited haplotypes determined in accordance with embodiments of the present invention.

In FIG. 23 shows different types of loci (alpha (A) and beta (B)) for maternal haplotypes relative to paternal haplotype according to embodiments of the present invention.

In FIG. 24 is a flowchart illustrating a method for identifying 2400 de novo mutations in the genome of an unborn fetus in a pregnant woman.

In FIG. 25A shows the absolute number and percentage of SNPs representing various combinations of paternal, maternal, and fetal (CVS) genotypes according to embodiments of the present invention.

In FIG. 25B is a table showing alignment statistics for the first 20 flow cells.

In FIG. 26 is a table showing fractional concentrations of fetal DNA calculated from SNPs by two methods according to embodiments of the present invention.

In FIG. 27A is a graph illustrating the observed percentage of SNPs in the subgroup in which the fetal allele is observed on the sequencing of the first 20 flow cells, and FIG. 27B is a plot of coverage vs. number of spans read according to embodiments of the present invention.

In FIG. 28A and B are graphs of the correlation between paternally inherited allele coverage and the number of mapped sequence reads and the number of sequenced flow cells, respectively, according to embodiments of the present invention.

In FIG. 29A shows the correlation between the false positive rate and the number of flow cells sequenced, and FIG. 29B shows the correlation between the false positive rate and the number of flow cells sequenced according to embodiments of the present invention.

In FIG. 30 shows the coverage of fetal-specific SNPs for different numbers of flow cells analyzed according to embodiments of the present invention.

In FIG. 31 shows the accuracy of Type A analysis using data from 10 flow cells according to embodiments of the present invention.

In FIG. 32 shows the accuracy of Type B analysis using data from 10 flow cells according to embodiments of the present invention.

In FIG. 33 shows the accuracy of Type A analysis using data from 20 flow cells according to embodiments of the present invention.

In FIG. 34 shows the accuracy of Type B analysis using data from 20 flow cells according to embodiments of the present invention.

In FIG. 35A and B show mutated and wild-type sequence stretches at codons 41/42 according to embodiments of the present invention.

In FIG. 36 is a type A RHDO analysis table, and FIG. 37 shows an analysis of RHDO type B according to embodiments of the present invention.

In FIG. 38A and B show the SPRT classification results for PW226 as an example.

In FIG. 39 is a table summarizing the RHDO analysis results for five cases according to embodiments of the present invention.

In FIG. 40 is a plot of sequencing completeness versus number of flow cells sequenced according to embodiments of the present invention.

In FIG. 41 is a graph of the size of fetal and total sequences for the entire genome, and FIG. 42A-C are similar plots for each chromosome individually according to embodiments of the present invention.

In FIG. 43 is a block diagram of an example computer system 4300 used with the system and methods according to embodiments of the present invention.

Definitions

The term biological sample in the present invention refers to any sample taken from a subject (eg, a human, such as a pregnant woman) and containing one or more nucleic acid molecules of interest.

The term nucleic acid or polynucleotide refers to deoxyribonucleic acid (DNA) or ribonucleic acid (RNA) and their polymers in both single and double stranded form. Unless otherwise stated, the term encompasses nucleic acids containing known analogs of naturally occurring nucleotides that have the same binding properties as a standard nucleic acid and are metabolized in the same manner as naturally occurring nucleotides. Unless otherwise stated, a particular nucleic acid sequence also includes, by implication, its conservatively modified variants (eg, degenerate codon substitutions), alleles, orthologs, SNPs, and complementary sequences, as well as sequences explicitly stated. In particular, substitutions for degenerate codons can be achieved by creating sequences in which one or more individual (or all) codons in the third position have been replaced with mixed bases and/or deoxynosine residues (Batzer et al., Nucleic Acid Res. 19: 5081 (1991); Ohtsuka et al., J. Biol. Chem. 260:2605-2608 (1985); and Rossolini et al., Mol. Cell. Probes 8:91-98 (1994)). The term nucleic acid is used interchangeably with gene, cDNA, mRNA, small non-coding RNA, microRNA (miRNA) interacting with Piwi-RNA, and short hairpin RNA (shRNA) encoded by a gene or locus.

The term gene refers to a segment of DNA involved in the production of a polypeptide chain or a transcribed RNA product. It may include regions preceding or following the coding region (leader or trailer), as well as intermediate sequences (introns) between individual coding segments (exons).

The term clinically relevant nucleic acid sequence (also referred to as target sequence or chromosome) in the present invention may refer to a polynucleotide sequence corresponding to a segment of a larger genomic sequence whose possible imbalance is being tested, or the larger genomic sequence itself. One example is the sequence of chromosome 21. Other examples include chromosomes 18, 13, X, and Y. Still other examples include mutated genetic sequences or genetic polymorphisms or copy number variations that a fetus can inherit from one or both parents or as a mutation de novo in the fetus. In some embodiments, a plurality of clinically relevant nucleic acid sequences or a plurality of markers of a clinically significant nucleic acid sequence may be used to provide data to detect an imbalance. For example, data from five non-contiguous sequences on chromosome 21 can be additively used to determine a possible disorder on chromosome 21, effectively reducing the sample volume required to 1/5.

The term based in the present invention means at least partially based on and refers to one value (or result) that is used to determine another value, as it occurs at the input of the method and at the output of the method. The term inference in the present invention also refers to the relationship between the input of a method and the result of that method, as occurs when inference means calculation by a formula.

The term parameter in the present invention denotes a numerical value characterizing a set of quantitative data and/or a quantitative relationship between sets of numerical data. For example, the parameter is the ratio (or ratio function) between the first amount of the first nucleic acid sequence and the second amount of the second nucleic acid sequence.

In the present invention, the term locus, or its plural form loci, refers to the location or address of any stretch of nucleotides (or base pairs) having variation between genomes.

The term sequence imbalance in the present invention means any significant deviation, defined by at least one cutoff value, in the amount of a clinically significant nucleic acid sequence from a standard amount. Sequence imbalances can include chromosomal imbalances, allelic imbalances, mutational imbalances, haplotype imbalances, and other similar imbalances. As an example, an allelic or mutational imbalance occurs when the genotype of the fetus differs from the maternal genotype, resulting in an imbalance at a particular locus in the sample.

The term chromosomal aneuploidy in the present invention means a change in the number of chromosomes from the diploid set in the genome. Change can be gain or loss. It may include an entire chromosome or a portion of a chromosome.

The term haplotype in the present invention refers to the combination of alleles at multiple loci that are passed together on the same chromosome or region of the chromosome. A haplotype can refer to just one pair of loci, or to a portion of a chromosome, or to an entire chromosome. The term alleles refers to alternative DNA sequences at the same physical locus of the genome, which may or may not result in different phenotypic traits. In any particular diploid organism with two copies of each chromosome (except for the sex chromosomes in males), the genotype for each gene contains a pair of alleles present at that locus that are the same in homozygotes and different in heterozygotes. A population of organisms or a species usually contains several alleles at each locus in different individuals. A genomic locus at which more than one allele is found in a population is called a polymorphic site. Allelic variation at a locus is measured by the number of alleles (i.e., the degree of polymorphism) or the proportion of heterozygotes (i.e., heterozygosity) in a population. In the present invention, the term polymorphism refers to any variation between individuals in the human genome, regardless of their frequency. Examples of such variations include, without limitation, single nucleotide polymorphisms, simple tandem repeat type polymorphisms, insertion-deletion type polymorphisms, mutations (which can cause disease), and copy number variations.

- 4 040939

Disclosure of the essence of the invention

Compilation of a partial genetic map or a complete genomic sequence of an unborn fetus can be carried out on the basis of haplotypes of polymorphic sequences of its parents. The term haplotype in the present invention refers to a combination of alleles at multiple loci that are transferred together on the same chromosome or chromosome region. For example, in embodiments, DNA fragments from a maternal sample (containing both maternal and fetal DNA) can be analyzed to identify alleles at certain predetermined loci (markers). You can then analyze the total number of DNA fragments of the corresponding alleles at these loci to determine the relative number of haplotypes at these loci and thereby determine which haplotypes are inherited by the fetus from the maternal and/or paternal genome. By identifying fetal haplotypes, the fetal genotype can be determined from individual loci in the appropriate genomic region, including target loci. In various embodiments, loci at which parents are represented by certain combinations of homozygotes and heterozygotes can be analyzed in such a way as to determine regions of the fetal genome. In one embodiment, control haplotypes representing common haplotypes in a population are used to determine maternal and paternal genomes, along with analysis of DNA fragments from a maternal sample.

An example of using one embodiment to determine at least a portion of the fetal genome would be to determine paternity by comparing the inferred genotype or haplotype of the fetus with the genotype or haplotype of the putative father. Another example is the detection of one or more de novo mutations acquired by the embryo, or the detection of cases of meiotic recombination that occurred during the formation of gametes from its parents. These are the gametes that have undergone fertilization, and a fetus has developed from the resulting zygote.

In addition, some embodiments also make it possible to determine the genomic sequence of an unborn fetus with the required resolution. For example, in some embodiments, the complete or near-complete genomic sequence of an unborn fetus can be determined. In one embodiment, the resolution of the fetal genomic sequence to be determined depends on knowledge of the genomes of the father and mother along with sequencing information of the maternal biological sample containing the fetal nucleic acids. In the case where the complete or nearly complete genomic sequences of the father and mother are known, it is possible to deduce the complete or almost complete genomic sequence of the unborn fetus.

In other embodiments, only the genomic sequences of selected regions of the genome are elucidated, for example, for prenatal diagnosis of selected genetic, epigenetic (such as imprinting disorders), or chromosomal disorders. Examples of genetic disorders to which an embodiment may apply include hemoglobinopathies (such as beta thalassemia, alpha thalassemia, sickle cell anemia, hemoglobin E disease), cystic fibrosis, and sex-related diseases (such as hemophilia and Duchenne muscular dystrophy). Additional examples of mutations that can be determined by embodiment can be found in Online Mendelian Inheritance in Man (www.ncbi.nlm.nih.gov/omim/getmorbid.cgi).

Some embodiments can also be used to determine the fractional concentration of fetal DNA, which can be performed without prior knowledge of the specific genomes of the parents. A similar analysis can also be applied to determine the completeness needed to accurately determine the fetal genome. So coverage can be used to estimate how much data needs to be analyzed to get accurate results.

I Introduction

When a maternal sample (such as plasma or serum) is used as a material for elucidating the haplotype of a fetus, two main problems can arise. The first problem is that maternal plasma or serum consists of a mixture of fetal and maternal DNA, with fetal DNA constituting a small population. The mean/median proportion of fetal DNA has been found to be between 5% and 10% of total maternal plasma DNA in the first two trimesters of pregnancy (Lo YMD et al., Am J Hum Genet 1998, 62: 768-775; Lun FMF et al ., Clin Chem 2008, 54: 1664-1672). Because DNA is released by maternal blood cells during blood clotting, the fractional concentration of fetal DNA in maternal serum may be even less than that in maternal plasma. Therefore, in some embodiments, maternal plasma is preferred over maternal serum.

The second problem is that fetal and maternal DNA in maternal plasma consist of short fragments (Chan KCA et al., Clin Chem 2004, 50: 88-92). Thus, in maternal plasma, fetal-derived DNA is usually shorter than maternal-derived DNA. Most fetal DNA in maternal plasma is less than 200 bp in length. Using such short DNA fragments in plasma alone would make it difficult to haplotype genetic polymorphisms over long stretches of the genome. The above concerns regarding maternal plasma and serum also apply to the detection of fetal DNA in maternal urine (Botezatu I et al., Clin Chem 2000, 46: 1078-1084). Fetal DNA makes up only a small fraction of the DNA in a pregnant woman's urine, and fetal DNA in a mother's urine also consists of short pieces of DNA.

- 5 040939

A. Sequencing and analysis of the maternal sample

An approach taken in some embodiments to deal with the first problem is to use a method that allows high precision quantitative genotyping of nucleic acids derived from a maternal biological sample. In one embodiment of this approach, accuracy is achieved by analyzing a large number (eg, millions or billions) of nucleic acid molecules. Moreover, accuracy can be improved by analyzing single nucleic acid molecules or by clonal amplification of single nucleic acid molecules. In one embodiment, massive parallel DNA sequencing is used, such as that performed on the Illumina Genome Analyzer platform (Bentley DR et al., Nature 2008, 456: 53-59), Roche 454 platform (Margulies M et al., Nature 2005, 437: 376-380), ABI SOLiD platform (McKernan KJ et al., Genome Res 2009, 19: 1527-1541), Helicos single molecule sequencing platform (Harris TD et al., Science 2008, 320: 106-109), sequencing in real time using single polymerase molecules (Science 2009, 323: 133-138) and nanopore sequencing (Clarke J et al., Nat Nanotechnol. 2009, 4: 265-70). In one embodiment, massive parallel sequencing is performed on a random subset of nucleic acid molecules in a biological sample.

In some embodiments, it may be advantageous to read as long a sequence as possible from each molecule. One limitation on the length of the sequence read may be the nature of the nucleic acid molecules in the maternal biological sample. For example, it is known that most DNA molecules in maternal plasma consist of short fragments (Chan CA et al., Clin Chem 2004, 50: 88-92). Moreover, the read length must be balanced by the robustness of the sequencing system for long read lengths. For some of the above systems, sequencing from both ends of the molecule, so-called paired-end sequencing, may be preferred. As an example, one approach is to sequence 50 bp. from each end of the DNA molecule, resulting in a sequence of 100 bp. per molecule. In another embodiment, sequenced at 75 bp. from each end of the DNA molecule, resulting in a sequence of 150 bp. per molecule.

After sequencing, the resulting sequences are aligned with the control human genome. Since embodiments detect genomic variations inherited from parents by the unborn fetus, the alignment algorithm must account for sequence variations. One example of such a software package is Illumina's Efficient LargeScale Alignment of Nucleotide Databases (ELAND). Another example of such a software package is SOAP (short oligonucleotide alignment program) and SOAP2 (Li R et al., Bioinformatics 2008, 24:713-714; Li R et al., Bioinformatics 2009 , 25:1966-1967).

The extent of DNA sequencing that may be required for this may depend on the resolution with which the fetal genetic map or fetal genomic sequencing is to be compiled. In general, the more molecules that are sequenced, the higher the resolution. Another determining factor in the resolution of the fetal genetic map or fetal genomic sequence at a given level or completeness of DNA sequencing is the fractional concentration of fetal DNA in the maternal biological sample. In general, the greater the fractional concentration of fetal DNA, the higher the resolution of the fetal genetic map or fetal genomic sequence that can be obtained at a given level of DNA sequencing. Since the fractional concentration of fetal DNA in maternal plasma is higher than in maternal serum, for some embodiments, maternal plasma is a more preferred type of maternal biological sample than maternal serum.

The performance of the above sequencing-based methods can be improved by indexing or barcoding. Thus, a sample- or patient-specific suffix or barcode can be added to the nucleic acid fragments in a particular sequencing nucleic acid library. Several of these libraries can then be mixed and sequenced together, each with its own sample or patient-specific index or barcode. After a sequencing reaction, sequencing data can be extracted from each sample or patient based on a barcode or index. Such a strategy can increase the productivity and thus the cost-effectiveness of embodiments of the present invention.

In one embodiment, nucleic acid molecules in a biological sample may be selected or fractionated prior to quantitative genotyping (eg, sequencing). In one embodiment, nucleic acid molecules are processed with a device (eg, microarray) that can preferentially bind nucleic acid molecules from selected loci in the genome (eg, a region on chromosome 7 containing the CFTR gene). Thereafter, sequencing can preferably be performed on the nucleic acid molecules captured by the device. This scheme allows you to target sequencing to a region of interest in the genome. One embodiment of such a scheme can use the Nimblegen Sequence Capture System (www.nimblegen.com/products/seqcap/index.html) or the Agilent SureSelect Target Enrichment System

- 6 040939

Target Enrichment System (www.opengenomics.com/SureSelect_Target_Enrichment_System) or similar platforms. In some embodiments, nucleic acid molecules from selected regions of the genome are subjected to randomized sequencing.

In another embodiment, a genomic region of interest in a biological sample may first be amplified with one or more sets of amplification primers. After that, quantitative genotyping can be carried out, for example, sequencing of the amplification products. One implementation of such a scheme may use the RainDance system (www.raindancetech.com/technology/pcr-genomics-research.asp). In some embodiments, the amplified nucleic acid molecules are subjected to randomized sequencing.

Nucleic acid molecules in a biological sample can also be size fractionated. Since fetal DNA is known to be shorter than maternal DNA in a maternal sample (Li et al., Clin Chem 2004, 50: 1002-1011; US Patent Application 20050164241; US Patent Application 20070202525), can be selected and then used for quantitative genotyping, for example, sequencing, the fraction of smaller molecules. This fraction will contain a higher proportion of fetal DNA compared to the original biological sample. Thus, sequencing a fraction enriched in fetal DNA makes it possible to map the fetus or derive a genomic sequence at a higher resolution at a given level of analysis (eg, sequencing completeness) than when using an unenriched sample. This will make the technology more cost-effective. As an example of size fractionation methods, one can use: (i) gel electrophoresis followed by extraction of nucleic acid molecules from certain parts of the gel; (ii) nucleic acid binding matrices with differential affinity for nucleic acid molecules of different sizes; or (iii) filtration systems with differential retention of nucleic acid molecules of different sizes.

And in another embodiment, nucleic acid molecules of a certain size or range of sizes are preferably subjected to analysis after nucleic acid sequencing. For example, you can perform sequencing at paired ends, in which sequencing occurs from both ends of the DNA molecule. The genomic coordinates of both of these ends can then be mapped to the reference human genome. The size of the molecule can then be calculated by subtracting the genomic coordinates of the two ends. One way to perform this paired-end sequencing is to use the Illumina Genome Analyzer's paired-end sequencing protocol. Another way to derive the size of a DNA molecule is to sequence the entire DNA molecule. This is most easily done with relatively long read span sequencing platforms such as the Roche 454 platform (Marguelis et al., Nature 2005, 437:376-380) and Pacific Biosciences Real-Time Single Molecule Technology (SMRT™) technology (Eid et al. ., Science 2009, 323: 133138). After calculating the size of the nucleic acid molecules, one can focus on the subsequent analysis of molecules below a certain cut-off, while increasing the fractional concentration of fetal DNA. Analysis of this subset of molecules makes it possible to generate a genetic map of the fetus or genomic sequences of the fetus with fewer analyzed molecules after size selection than without such a procedure. In one embodiment, a 300 bp cutoff is used. In other embodiments, a cutoff of 250 bp, 200 bp, 180 bp, 150 bp, 125 bp, 100 bp may be used. or 75 bp

B. Using parental genomes as a scaffold

To address the second problem, in some embodiments, maternal chromosome haplotypes may be used as a scaffold. And paternal chromosome haplotypes can be used as another scaffold. This scaffold can be compared to the fetal genetic information obtained from a maternal sample containing the fetal DNA. Such fetal genetic information can be used to determine how the mother's and/or father's scaffold was unfolded in the fetal genome, while using the constituent parts of the scaffold to determine the nascent fetal genome.

Parental haplotypes can be composed of the genomic DNA of the father and mother and other family members, such as the brother or sister of the fetus in the current pregnancy. It is possible that the availability of parental haplotypes will become increasingly common due to the declining cost of genome sequencing. In one scenario, if the genomes of one or both parents have already been sequenced and their haplotypes in one or more regions of the chromosomes have been determined, then this information can be used as the above framework.

Any genotyping platforms known in the art that are capable of detecting sequence variation in the genome can be used, including DNA sequencing, microarrays, hybridization probes, fluorescence methods, optical methods, molecular barcodes, and single molecule imaging (Geiss GK et al., Nat Biotechnol 2008, 26: 317325), single molecule analysis, PCR, digital PCR, mass spectrometry (such as the Sequenom MassARRAY platform), etc. As an extreme example, it is possible to determine the DNA sequence of the father and mother by sequencing the entire genome using a massively parallel sequencing method (for example , Bentley DR et al., Nature 2008, 456: 53-59 McKernan KJ et al., Genome Res 2009, 19: 1527-7 040939

1541) . An example of sequence variations that may be of interest are single nucleotide polymorphisms (SNPs). A particularly preferred way to determine parental genotypes is by analysis of SNPs on microarrays throughout the genome or at specific regions of the genome, for example, those containing genes whose mutations can cause genetic diseases (such as genes in the beta-globin cluster or the cystic fibrosis-causing transmembrane conductance regulator (CFTR) gene) ). In addition to sequence variations, copy number variations can also be used. Sequence variations and copy number variations are referred to as polymorphic genetic traits (PMFs).

In one aspect, maternal genotypes on chromosomes or regions of chromosomes of interest can be compiled into haplotypes. This can be done by analyzing other family members related to the mother, such as the mother's son or daughter, parents, siblings, and the like. Haplotypes can also be compiled by other methods, which are well known in this field and are listed above.

The genotype information can then be extended to parental haplotype information by comparing the genotype information with other family members, such as siblings of the current pregnancy's fetus or grandparents' genotypes, etc. Parental haplotypes can also be compiled by other methods well known to those skilled in the art. Examples of such methods include single molecule based methods such as digital PCR (Ding C and Cantor CR, Proc Natl Acad Sci USA 2003, 100: 7449-7453; Ruano G et al., Proc Natl Acad Sci USA 1990, 87: 6296-6300), sperm haplotyping (Lien S et al., Curr Protoc Hum Genet 2002, Chapter 1 :Unit 1.6) and imaging techniques (Xiao M et al., Hum Mutat 2007, 28: 913-921). Other methods include allele-specific PCR based methods (Michalatos-Beloin S et al., Nucleic Acids Res 1996, 24: 4841-4843; Lo YMD et al., Nucleic Acids Res 1991, 19:3561-3567), cloning, and cleavage by restriction enzymes (Smirnova AS et al., Immunogenetics 2007, 59: 93-8), etc. Still other methods are based on the distribution and structure of the non-equilibrium linkage of haplotype blocks in the population, which makes it possible to compile the maternal haplotype according to statistical estimates (Clark AG, Mol Biol Evol 1990, 7:111-22; 10:13-9; Salem RM et al., Hum Genomics 2005; 2:39-66).

C. Use of genome information from a maternal sample to construct a scaffold

In one embodiment, the Haplotype Relative Dose (RHDO) method is used to find out which maternal chromosome has been transferred to the fetus. The basic principle of this method follows from the example when the mother is heterozygous for each of the genetic polymorphisms. So, there are two haplotypes and the relative dose of these haplotypes should be 1:1. However, the presence of a small proportion of fetal DNA in the maternal sample can change the relative dose of the haplotype. This is because the fetus inherits half of its haplotype set from the mother and the other half from the father. Moreover, for each chromosome, the fetus could inherit a mosaic of haplotypes derived from one or another homologous chromosome from each parent, depending on the presence of meiotic recombination. All these factors can cause the relative dose of haplotypes to deviate from the 1:1 ratio in maternal DNA. Thus, for a given chromosome or region of a chromosome, the constituent alleles of these haplotypes can be retrieved from analytical data (eg, sequencing data) obtained from a maternal sample.

A statistical procedure can then be performed to determine the relative dose of haplotypes, or whether one of the haplotypes is more strongly represented than the other.

The classification threshold for this statistical procedure can be set depending on the fractional concentration of fetal DNA. In general, a higher fractional concentration of fetal DNA contributes to reaching the threshold with fewer molecules. The classification threshold can also be set depending on the number of successfully classified fragments that need to be obtained for the entire genome or for the desired genome regions. In one embodiment, a sequential probability ratio test (SPRT) may be applied.

In one embodiment, Relative Mutation Dose (RMD) can be used to determine the relative allele dose for certain maternal polymorphisms, as described in US Patent Application 2009/0087847. These relative numbers can be used to determine the haplotype in a fetus (eg, when polymorphisms occur at successive or related loci). One implementation of this targeted approach uses polymerase chain reaction (PCR) to amplify specific sequences from distinct portions of the genome for RMD analysis. In order for this approach to determine fetal inheritance by RMD to be extended to a large region of the genome or the entire genome, a large amount of maternal sample is required.

In one embodiment, using randomized sequencing, regions of the genome of interest are not targeted. Thus, the number of sequences obtained in the regions of interest in the genome may not be as large as with a targeted approach (unless very deep sequencing is carried out). However, it is possible to combine outcomes for multiple linked polymorphisms to achieve the necessary statistical significance for diagnostic purposes. The practical significance of this embodiment of sequencing is that

- 8 040939 can save costs by eliminating the need for too deep sequencing. In addition, less maternal sample needs to be applied than with digital PCR approaches.

In addition, it may be desirable to perform such RHDO analysis in blocks. In other words, each chromosome can be analyzed in one or preferably more than one block. In one aspect, the latter makes it possible to observe meiotic recombination. For example, it may be that the haplotype of one segment of a particular fetal chromosome comes from one of the maternal homologous chromosomes, while another segment of the same fetal chromosome has a haplotype from another homologous chromosome. An SPRT analysis can facilitate this segmentation.

For example, SPRT analysis can be performed on adjacent SNPs exhibiting the desired parental genotype configuration (ie homozygous father and heterozygous mother) starting at one end of the chromosome. This should continue until the SPRT analysis shows that one of the maternal haplotypes dominates the maternal plasma analysis (eg, sequencing data). The SPRT analysis can then be readjusted and restarted from the next adjacent SNP showing the desired parental genotype configuration. This can be continued again until the SPRT analysis shows once again that one of the maternal haplotypes predominates in maternal plasma analyzes (eg, sequencing data). This process may continue until the last selected SNP on a given chromosome. These diverse SPRT haplotype segments on a chromosome can then be compared to the haplotypes of two homologous chromosomes in the mother's genome. Meiotic recombination is noted when it turns out that haplotype segments in the fetus have switched from one maternal homologous chromosome to another. This system works even if there is more than one homologous recombination per chromosome.

As will be described later, RHDO analysis can also be carried out in those regions of the genome in which both the father and mother are both heterozygous for their genetic polymorphisms. This scenario is particularly useful in cases where the father and mother both contain a mutated copy of the disease-causing gene from the same ancestor, such as when they are closely related or when the disease-causing mutation is predominant due to a strong founder effect (i.e. most individuals with this mutation inherited the same haplotype from a common founding ancestor of the population). Thus, the paternal and maternal haplotypes at this site can be used to calculate the fetal haplotype.

II. Compilation of the fetal genome from the mother's genome

The following describes the compilation of a genetic map of the fetus or the elucidation of the genomic sequence of the fetus with explicit knowledge of the parental genomes.

A. Method

In FIG. 1 is a flow diagram of a method for determining 100 at least a portion of the genome of an unborn fetus of a pregnant woman. The fetus has a father and a pregnant mother. The father has a paternal genome with two haplotypes, and the mother has a maternal genome with two haplotypes. Method 100 analyzes nucleic acid molecules (fragments) from a biological sample taken from a pregnant woman to determine the fetal genome. Method 100 is described primarily with the father homozygous and the mother heterozygous at multiple loci, while other examples describe other embodiments.

Method 100, and any of the methods described herein, may be performed in whole or in part by a computer system including a processor that can be configured to perform the steps. Thus, embodiments are directed to computer systems configured to perform the steps of any of the methods described herein, potentially with different components performing the respective step or respective group of steps. Although they are presented as numbered steps, the steps of the methods may be performed simultaneously or in a different order. Moreover, parts of these steps can be used in conjunction with parts of other steps from other methods. In addition, all steps or parts of the steps may be optional. In addition, any steps of any of the methods may be performed using modules, loops, or other means to perform these steps.

At step 110, a first set of loci is determined for which the maternal genome is heterozygous. In one embodiment, this determination may be performed as part of genotyping of the father and mother at the genome level or at selected genomic loci. In other embodiments, the determination of the first set of loci may be carried out during the analysis of the maternal sample, which is described in the following sections.

At step 120, each of the two maternal haplotypes spanning the first set of loci is determined. As stated above, the maternal genome can be obtained by direct sequencing. In other embodiments, genotyping can be performed at multiple loci, followed by a mapped genome of an individual who should have a similar genome, such as a family member, or a control genome common to the same or similar population. In one embodiment, step 120 may be performed on all or part of the maternal genome first, and then the maternal

- 9 040939 the genome can be examined to identify loci for which the mother is heterozygous.

In one aspect, compiling a paternal chromosome haplotype is optional. However, if paternal haplotypes can be compiled, additional information can be obtained from the sequencing results. Such additional information includes the possibility of performing haplotype relative dose analysis for regions for which both parents are heterozygous. Other additional information that can be obtained if a paternal haplotype is available is information regarding meiotic recombination involving one or more paternal chromosomes to establish whether disease-causing alleles associated with such polymorphisms are passed on to the fetus.

At step 130, the allele inherited by the fetus from the father is determined at each of the first set of loci. In some embodiments, genomic loci are used for which the father is homozygous and the mother is heterozygous (as indicated in step 110). Thus, if the father is homozygous for these loci, then the alleles inherited from the father will be known. Genotyping of the father to determine the loci for which the father is homozygous may be performed by any of the methods described herein. In one embodiment, the determination of the first set of loci may be based on the genotyping of the father and mother to find the loci for which the father is homozygous and the mother is heterozygous.

In another embodiment, the second set of loci of the father's genome for which he is heterozygous can be used to determine the paternal haplotype inherited by the fetus from the first set of loci for which the father is homozygous. For example, if the maternal genome is homozygous for the second set of loci, then alleles present in the paternal genome at the corresponding loci from the second set of loci and not present in the maternal genome can be identified. The paternally inherited haplotype can then be identified as the haplotype with identified alleles and used to determine paternally inherited alleles at the first set of loci. These aspects of paternal haplotype determination are discussed in more detail below.

In step 140, a plurality of nucleic acid molecules are analyzed from a biological sample taken from a pregnant woman. The sample contains a mixture of maternal and fetal nucleic acids. A maternal biological sample may be taken and then processed for analysis. In one embodiment, maternal plasma and serum are used. In other embodiments, maternal blood, maternal urine, maternal saliva, uterine lavage, or fetal cells derived from maternal blood may be used.

In one embodiment, the analysis of nucleic acid molecules includes determining the location of the nucleic acid molecule in the human genome and determining its allele at an individual locus. Thus, in one embodiment, quantitative genotyping can be performed using established alleles of nucleic acid molecules from the same locus. Any suitable method can be used to determine the genomic location and alleles (genotype information) of nucleic acid molecules in a maternal sample. Some of these methods are described in US Pat. Nos. 12/178,181 and 12/614,350 and the application titled Size Based Genome Analysis.

At step 150, based on the established alleles of the nucleic acid molecules, the numbers of the corresponding alleles at each of the first set of loci are determined. In one embodiment, these numbers may be the number of alleles of each type at the first locus. For example, six A and four T. In another embodiment, the number can mean the size distribution of nucleic acid molecules containing certain alleles. For example, the relative number can also include the size distribution of fragments with a certain genotype, which can mean the relative number of fragments of a certain length. Such relative numbers can also provide information about what genotype the fetal genome has, since the fragments in the fetus are usually smaller than those of the mother. Some examples of quantities and methods are described in US applications 12/178181 and 12/614350 and the application titled Size Based Genome Analysis.

In one embodiment, the relative numbers of alleles at one locus can provide information about which genotype is inherited by the fetus (eg, after the data set reaches sufficient statistical significance). For example, relative amounts can be used to determine that there is a sequence imbalance relative to the mother's genotype at one locus. The above related patent applications provide exemplary embodiments for detecting sequence imbalance at a particular locus or region.

In step 160, the relative amounts of corresponding alleles of nucleic acid molecules at more than one locus of the first plurality of loci are compared. In some embodiments, the amount of each allele across all loci of the first set of loci constituting the haplotypes is summed before comparison is made. The total numbers of parental haplotypes are then compared to determine if the haplotype is overrepresented, equally overrepresented, or underrepresented. In other embodiments, the numbers of alleles at loci are compared and the comparison is made across multiple loci. For example, you can sum partition values (such as differences or ratios) that can be used when comparing against a cutoff limit. Each of these embodiments is applicable to any of the comparison operations described here.

In various embodiments, the relative amount may mean the occurrence of each

- 10 040939 a fragment with a specific allele at a specific locus, the occurrence of fragments from any locus (or any loci in the site) for a specific haplotype, and a statistical value of occurrence (for example, average) for a specific locus or for a specific haplotype. Thus, in one embodiment, comparison may mean determining the amount of separation (eg, difference or ratio) of one allele compared to another allele at each locus.

At step 170, based on the comparison, it is possible to determine the haplotype inherited by the unborn fetus from the mother in the part of the genome covered by the first set of loci. In one embodiment, in order to determine which of the maternal chromosomes has passed to the fetus, the method of relative haplotypes (RHDO), for example, as described above, is used. Since the mother is heterozygous for each of the first loci, the first loci correspond to two haplotypes for the genome region of the first loci. The relative dose of these haplotypes should be 1:1 if the sample was from the mother only. Deviation or lack of deviation from this ratio can be used to determine the haplotype of the fetus inherited from the mother (and from the father, as described in more detail below). Thus, for a given chromosome or region of the chromosome, the alleles that make up these haplotypes can be found from analytical data (for example , sequencing data) obtained in step 130.

Since many loci are analyzed and compared with the mother's haplotype, the sequences between the loci can be attributed to a particular haplotype. In one embodiment, if multiple loci correspond to a certain haplotype, then it can be concluded that the sequence segments between the loci will be the same as the maternal haplotype. Due to the phenomenon of meiotic recombination, the final haplotype inherited by the fetus may consist of a mosaic of haplotype segments derived from one of these two homologous chromosomes. Embodiments may exhibit such recombination.

The resolution with which such recombination can be detected depends on the number and distribution of genetic markers that were identified in the DNA of the mother and father, and on the threshold that is applied in the subsequent bioinformatics analysis (for example, by SPRT). For example, if the comparison indicates that the allele inherited from the mother at each of the first set of consecutive loci corresponds to the first haplotype, then the first haplotype is considered to be inherited at the position in the genome corresponding to the first set of loci. If the second set of consecutive loci indicates that the second haplotype is inherited, then the second haplotype is considered to be inherited at the position in the genome corresponding to the second set of loci.

In one embodiment, since multiple loci are analyzed, the haplotype can be determined with greater accuracy. For example, statistics for some loci may not be decisive, but in combination with statistics for other loci, it can be established which haplotype is inherited. In another embodiment, each locus can be analyzed independently to generate a classification, and then these classifications can be analyzed to determine which haplotype is inherited for that region.

In one embodiment, a statistical procedure can be performed to determine the relative dose of a haplotype (eg, if one of the haplotypes is overrepresented before another haplotype). The classification threshold for this statistical procedure can be set depending on the fractional concentration of fetal DNA. In general, a higher fractional concentration of fetal DNA allows the threshold to be reached with fewer molecules. The classification threshold can also be set depending on the number of successfully classified segments that need to be obtained for the entire genome or individual regions of the genome.

Referring again to FIG. 1, at step 180, the fetal genome can be analyzed for mutations. For example, embodiments can be used to search for mutations that cause genetic diseases in a particular population. Examples of mutations that can be identified using embodiments can be found on the Online Mendelian Inheritance in Man website (www.ncbi.nlm.nih.gov/omim/getmorbid.cgi). These mutations can be looked for in steps 140-160 or as a separate step as described below. For example, searching for mutations in families in which the father is a carrier of one or more mutations that are absent from the mother can be conducted on analytical data (eg, sequencing data) from the maternal biological sample.

In addition to identifying existing mutations, one can search for polymorphic genetic markers linked to a mutant or wild-type allele in the father or mother. For example, when analyzing RHDO, it may turn out that the fetus inherited from the mother a known haplotype that carries a mutation that causes the disease. Embodiments of the invention can also be used for non-invasive prenatal diagnosis of diseases caused by deletions of portions of chromosomes, such as Southeast Asian deletion causing alpha thalassemia. In the case when both the father and mother are carriers of the deletion, and the fetus is homozygous for the deletion, then during massive parallel DNA sequencing from the mother's plasma, a decrease in the frequency of DNA sequences originating from the deleted site in the maternal plasma should be observed.

B. Example

This section describes an example of embodiments (eg, method 100) applicable to a single nucleotide polymorphism (SNP) in which the mother is heterozygous. SNP alleles on the same chromosome form a haplotype, with the mother having a homologous pair of each chromosome, i.e. two haplotypes. To illustrate how such a determination is made, consider a segment on chromosome 3, for example, as shown in FIG. 2.

In FIG. 2 presents two haplotypes for the father and two haplotypes for the mother for a certain segment of the genomic code corresponding to them. Within this segment, five SNPs were found for which the father is homozygous and the mother is heterozygous for all 5 of these SNPs. Two homologous chromosomes of the father have the same haplotype (Har), i.e. A-G-A-A-G (from top to bottom in Fig. 2). For simplicity, the paternal haplotypes are referred to as Hap I and Hap II, given that both are identical in this set of 5 SNPs. The mother had two haplotypes, namely Hap III - A-A-A-G-G and Hap IV - G-G-G-A-A.

In this example, SNPs can be further classified into two types. In FIG. 3 shows two types of SNPs according to embodiments of the present invention. Type A consists of those SNPs whose parental alleles are the same as in the maternal haplotype III. Type B consists of those SNPs in which the parental alleles are the same as in the maternal haplotype IV.

These two types of SNPs may require slightly different mathematical processing. Thus, in type A scenario, fetal inheritance of haplotype III would result in an overrepresentation of haplotype III relative to haplotype IV in maternal plasma (Fig. 4A). For example, considering only one SNP 410 for simplification, the A allele is inherited from the father, and if Hap III is inherited from the mother, then the fetus will have two A alleles in the sample, which will lead to overrepresentation of A. If the fetus inherited haplotype IV, then the excess representation will not be observed, since the fetus will also be heterozygous for A and G at this locus.

On the other hand, in scenario type B, fetal inheritance of haplotype III would result in equal representation of haplotype III and haplotype IV in maternal plasma (Fig. 4B). For example, considering SNP 420, inheritance of G from the father and A in Hap III would result in the fetus having equal amounts of A and G on SNP 420, just like the mother. If the fetus has inherited haplotype IV, then there will be overrepresentation, as is clear from the discussion above.

In FIG. 5A and B present an analysis by comparing the relative numbers (eg, occurrence) of fragments for each locus and whether the result of the comparison would indicate that a particular haplotype is inherited or not. For this example, any location in the genome that has a SNP that matches one of these paternal and maternal genotype configurations (eg, type A or type B scenarios) can be used. From maternal plasma sequencing data, one can focus on how many sequenced molecules correspond to a particular SNP allele. An SPRT analysis (or other comparison method) can be used to establish the presence of any allelic imbalance between these alleles (Lo YD et a., Proc Natl Acad Sci USA 2007, 104: 13116-13121).

In FIG. 5A shows an analysis of type A SNPs. As can be seen, for each SNP, comparison by SPRT of relative amounts (eg, determined by split size) with a cut-off value provides classification. In one embodiment, if the classification threshold for SPRT is reached, then it can be concluded that the fetus has inherited a certain maternal haplotype. You can then reconfigure the settings for the SPRT analysis. Then, one can proceed to the analysis of the adjacent SNP corresponding to the required genotype configuration, both in the direction from telomere to centromere, and vice versa; and you can start a new SPRT analysis for that SNP.

On the other hand, in one embodiment, if the SNP no SPRT classification has not been achieved, then one can again proceed to the neighboring SNP in the same way, except that the scores for the next SNP can be added to the previous one, and then the SPRT analysis can be performed again. This process can continue until the classification threshold is reached. In FIG. 5A and B show the execution of this process for Type A and Type B analyses. In one embodiment, all classifications are analyzed together to obtain a final classification for a site. For example, if a classification is obtained for the first group of SNPs and for the next group of SNPs, then both classifications can be compared to see if they agree.

In FIG. 6 shows the effect of changing the odds ratio in SPRT classification (Zhou W et al., Nat Biotechnol 2001, 19:78-81; Karoui NE et al., Statist Med 2006, 25:3124-33). In general, a lower classification probability ratio, such as 8, may make it easier to classify. This may increase the number of classified regions within the genome. However, it is to be expected that some of these sites will be misclassified. On the other hand, with a larger probability ratio for classification, for example 1200, classification will only become possible if more SNPs are evaluated. In this case, the number of classified regions within the genome may decrease. It should be expected that the number and proportion of misclassified parcels will be less compared to using a lower classification threshold.

In one embodiment, a classification is only scored if two consecutive SPRT classifications result in the same haplotype (referred to as the two consecutive block algorithm). In one aspect, a two-block algorithm can improve classification accuracy. In some embodiments, for any portion of the sequence, an SPRT for type A SNPs may be performed first, followed by another SPRT for type B SNPs. alternating groups of genetic landmarks (eg, SNPs). In implementations using the two-block algorithm, the two blocks may be of different types.

The results of the SPRT analysis for type A and type B should be checked for consistency or inconsistency in the classification results. To improve classification accuracy in a single embodiment (interleaved approach), a classification is only valid if type A and type B analyzes for a given genome region yield consistent results. If these two analyzes give conflicting results, then the results of the classification of the two adjacent regions following this, one on the centromere side and the other on the telomere side, should be checked. If these two adjacent sites give consistent results, then the first site can be classified as having the same haplotype with these two sites. If two adjacent sections do not give consistent results, then you can move on to the next two adjacent sections until consistency is observed. One variation on this theme is to move in only one direction and take the classification results of the next one, two or more adjacent parcels as the results of the original parcel under consideration. The general principle is to use the results of the classification of adjacent regions of the genome to check the results of the classification of this region.

III. Determination of paternal alleles inherited by the fetus

In FIG. 7 is a flow diagram of a method for determining 700 at least the paternally inherited portion of the genome in an unborn fetus of a pregnant woman. Method 700 analyzes nucleic acid molecules (fragments) from a biological sample taken from a pregnant woman to determine the fetal genome. The sample contains a mixture of maternal and fetal nucleic acids.

In step 710, each of the plurality of nucleic acid molecules from the biological sample is analyzed to determine the location of the nucleic acid molecule in the human genome and determine the allelic type of the nucleic acid molecule. Thus, in one embodiment, the genotypes of nucleic acid molecules at a specific location (locus) can be determined. For this analysis, you can apply any of the methods described above and throughout the text.

At step 720, a first set of loci is determined for which the paternal genome is heterozygous and the maternal genome is homozygous. In one embodiment, the first set of loci is obtained by determining the paternal and maternal genomes. These genomes can be tested for genomic loci for which the father is heterozygous and the mother is homozygous.

At step 730, the haplotype inherited by the unborn fetus from the father in that part of the genome covered by the first set of loci is determined based on the established genotypes at the first set of loci. In one embodiment, from analytical data (eg, sequencing data) at each of these loci, an allele is determined that is present in the father but not present in the mother's genome. The combination of these alleles will indicate the chromosome haplotypes that the fetus has inherited from the father.

In another embodiment, if the haplotypes of each of the chromosomes or desired chromosome regions in the father's genome are known, then it is possible to determine where meiotic recombination occurred in the father during spermatogenesis. So, meiotic recombination in the father is noted when the haplotype of the DNA segment in the chromosome inherited from the father differs between the fetus and the father. The inclusion of such recombination information may be useful when using analytical data (eg, sequencing data) for prenatal diagnosis of genetic diseases by linkage to genetic polymorphisms.

IV. Father and mother are heterozygous for a region of the genome

Embodiments provide for a scenario in which the father and mother are heterozygous in a region of the genome. This scenario may be particularly important in families where the father and mother are close relatives, and when the disease is associated with a predominant mutation due to a strong founding ancestor effect. Under such circumstances, it would be expected that if the father and mother of the unborn fetus are carriers of the mutant gene, then the haplotype of the chromosome carrying the mutant copy of the gene will be almost identical, except in cases of meiotic recombination. This type of analysis may be particularly useful in autosomal recessive diseases such as cystic fibrosis, beta thalassemia, sickle cell anemia, and hemoglobin E disease.

In FIG. 8 is a flowchart of a method for determining 800 at least a portion of the genome of an unborn fetus at a site in which the father and mother are heterozygous according to embodiments of the present invention.

At step 810, a first set of loci is determined for which both the father and mother are heterozygous. In one embodiment, the first loci can be determined by any of the methods described here. For example, you can sequence the entire paternal genome or parts of it, or genotype different parts

- 13 040939 to identify the first loci. In this case, each of the two paternal and each of the two maternal haplotypes for the first set of loci will become known.

As an example, in FIG. 9 shows the haplotypes of the father and mother, which are both heterozygous in a certain region of the genome. As you can see, both parents contain the mutant gene (allele) in region 1. In particular, Hap I of the father and Hap III of the mother contain the mutant gene. It is also seen that both father and mother may contain a different copy of the chromosome carrying a copy of the wild-type gene. In particular, the father's Hap II and the mother's Hap IV contain the wild-type gene. Thus, this example relates to determining whether a fetus has inherited a mutated gene. Chromosomes from the father and mother that carry the wild-type gene have the same haplotype in the immediate vicinity of the gene, but may have different haplotypes at a greater distance from the gene. Since this chromosome most likely comes from different ancestors, it is unlikely to have identical haplotypes between father and mother along the entire length of the chromosome.

At step 820, a second set of loci is determined for which the father is heterozygous and the mother is homozygous. As you can see, the first and second set of loci are on the same chromosome. Plot 2 presents such second loci. Site 2 is chosen such that the father is heterozygous for one or more SNPs at that site and the mother is homozygous at that site.

At step 830, fragments from the sample of the pregnant woman are analyzed to establish localization in the human genome and genotype. Localization can be used to determine that a fragment (nucleic acid molecule) includes one or more first loci or one or more second loci. This information can then be used to determine the haplotype inherited from the father and the haplotype inherited from the mother.

At step 840, it is determined which of the two parental haplotypes was inherited by the fetus by analyzing the established genotypes of multiple nucleic acid molecules from the biological sample at at least one of the second loci. For example, one could search for SNP alleles exclusively present in the paternal genome but absent from the maternal genome, such as the T allele denoted * and the A allele denoted ⁺ in FIG. 9 from analysis data (eg, location and genotype ascertained at 710) of the maternal biological sample. Just as is done in method 700, if the T allele indicated by * is detected in maternal plasma, then this means that haplotype II (Hap II) is inherited from the father by the fetus. Conversely, if the A allele, denoted ⁺ , is detected in maternal plasma, this means that Hap I is inherited from the father by the fetus.

In step 850, the relative amounts of the established genotypes of nucleic acid molecules at more than one of the first plurality of loci are compared. In one embodiment, the numbers across all loci are summed and the relative numbers of maternal haplotypes are compared. Relative numbers can mean occurrence, size distribution, and any other parameters that provide information about which genotype is present in the fetal genome at a particular locus.

At step 860, based on which haplotype is inherited from the father by the fetus and based on a comparison of relative amounts, the haplotype inherited by the unborn fetus from the mother in that part of the genome covered by the first set of loci is determined. Thus, to determine which of the two maternal haplotypes is inherited by the fetus, an RHDO analysis (e.g., as described above) of the SNP at site 1 from maternal biological sample analysis is performed, given the paternal haplotype inherited by the fetus at site 2. In one embodiment, it is contemplated that there was no recombination between regions 1 and 2 when these regions passed from parents to the fetus.

For example, consider a scenario where analysis of site 2 found that the fetus inherited Hap I from the father. The fetal inheritance of Hap III (which is identical to Hap I at site 1) from the mother would then result in an over-representation of Hap III relative to Hap IV in maternal plasma. Conversely, if the fetus has inherited Hap IV from the mother, then an equal representation of Hap III and Hap IV will be observed in maternal plasma.

As another example, consider a scenario where analysis of site 2 found that the fetus had inherited Hap II from the father. The fetal inheritance of Hap IV (which is identical to Hap II at site 1) from the mother would then result in an overrepresentation of Hap IV relative to Hap III in maternal plasma. Conversely, if the fetus has inherited Hap III from the mother, then an equal representation of Hap III and Hap IV will be observed in maternal plasma.

In the previous sections, we derived the fetal genome and fractional concentration of fetal DNA using data obtained from DNA sequencing from maternal plasma, as well as information about the genotype of the fetal parents. The following sections describe embodiments for deriving a fractional concentration of fetal DNA and fetal genotype without prior knowledge of maternal and paternal genotypes/haplotypes.

V. Determination of the fractional concentration of fetal DNA

In some embodiments, an optional step is to determine the fractional concentration of fetal DNA. In various aspects, this fractional concentration may determine the amount of analysis (eg, the amount of sequencing required) or determine the accuracy of the analysis for a given amount of data (eg, genome sequence completeness). The determination of the fractional concentration of fetal DNA can also be used to determine the cut-off limit in classifying which haplotype and/or genotype is inherited.

In one embodiment, the fractional concentration of fetal DNA is determined by extracting analytical data (eg, obtained at steps 140 and 710) for loci for which both the father and mother are homozygous, but with different alleles. For example, for an SNP with two alleles, A and G, the father may have AA and the mother GG, and vice versa. For such loci, the fetus will be an obligate heterozygote. In the example above, the fetus will have an AG genotype, and the proportion of the A allele in the maternal sample can be used to determine the fractional concentration of fetal DNA. In another embodiment, statistical analysis may be performed to determine the locus at which the mother is homozygous and the fetus is heterozygous. It does not require any information about the genome of the mother or the genome of the father.

As an alternative to extracting analytical data, the fractional concentration of fetal DNA can also be determined using another approach, such as PCR, digital PCR, or mass spectrometry-based methods using a set of polymorphic genetic markers (Lun FMF et al., Clin Chem 2008 , 54: 1664-1672). Another alternative is to use one or more genomic loci that differ in DNA methylation between fetus and mother (Poon LLM et al., Clin Chem 2002, 48: 35-41; Chan KCA et al., Clin Chem 2006, 52: 2211-2218 ; US Patent 6,927,028). Another alternative is to use the approximate fractional concentration of fetal DNA determined in a control population, for example, at a similar gestational age. However, since the fractional concentration of fetal DNA may differ from sample to sample, this latter approach can be expected to be less accurate than measuring the concentration specifically for the sample being tested.

A. Determination of fractional concentration in obligate heterozygotes

In those embodiments in which the fetus is an obligate heterozygote, the fractional concentration of fetal DNA can be determined using the following series of calculations (eg, massive parallel sequencing). Let p denote the number of alleles in the fetus that are absent from the maternal genome. Let q denote the amount of another allele, namely the allele that is in the genome of both the mother and the fetus. The fractional concentration of fetal DNA is given by the following equation ip

R + H

In one embodiment, such a calculation is performed on aggregate data for all polymorphic genetic loci or polymorphic genetic traits corresponding to the configuration of the parental genotypes (eg, both parents are homozygous, but for different alleles).

B. Determination by Informative SNPs

The fractional concentration of fetal DNA can also be determined for any locus for which the mother is homozygous and the fetus is heterozygous, not just when the mother is homozygous for one allele and the father is homozygous for another allele. Both methods assume that the locus is informative. The term informative SNP can be used in different contexts depending on what kind of information is needed. In one context, this is information about an allele in the fetal genome at a particular locus where the allele is not present in the maternal genome. Thus, the subset of SNPs for which the mother is homozygous and the fetus is heterozygous can be called informative SNPs in the context of determining the concentration of fetal DNA. Cases where both mother and fetus are heterozygous but differ in at least one allele can also be used as informative SNPs. However, triallelic SNPs are relatively rare in the genome.

In FIG. 10 is a flow diagram of a method for determining the 1000 fractional concentration of fetal material in a maternal sample, in accordance with embodiments of the present invention. In step 1010, fragments from a sample of a pregnant woman are analyzed to determine the location in the human genome and the type of alleles (which can lead to the determination of the genotype at this site). In one embodiment, the fragments are analyzed by sequencing multiple nucleic acid molecules from a biological sample taken from a pregnant woman. In other embodiments, real-time PCR or digital PCR may be used.

At step 1020, it is determined that one or more of the first loci are informative. In some embodiments, the maternal genome is homozygous, but a non-maternal allele is detected at the informative locus in the sample. In one embodiment, the fetal genome is heterozygous at all first loci and the maternal genome is homozygous at all first loci. For example, at one first locus, the fetal genome may contain the first and second alleles (eg, TA), and the maternal genome may contain two corresponding second alleles (eg, AA) at the first locus. However, such loci may not be known in advance, for example in situations where the fetus is not an obligate heterozygote.

In one embodiment, those SNPs for which the mother is homozygous are considered to determine the informative locus. For those SNPs for which the mother is homozygous, the fetus will either be homozygous for the same allele or heterozygous. For example, if the SNP is polymorphic in both A and T, and the mother has an AA genotype, then the fetus will have an AA or TA genotype. In this case, the presence of the T allele in the maternal plasma sample means that the fetus will have the TA genotype, not AA. Some embodiments target the extent to which the presence of the T allele signifies the TA genotype by calculating the required cutoff, as described below.

At step 1030, for at least one of the first loci, a first number p of occurrences (occurrence) of the corresponding first allele and a second number q of occurrences (occurrence) of the corresponding second allele are determined. In one embodiment, the number of fetal-specific (T allele) and shared (A allele) alleles in maternal plasma is determined by various methods, such as, but not limited to, real-time PCR, digital PCR, and massive parallel sequencing.

At 1040, a fractional concentration is calculated based on the first and second numbers. In one embodiment, for a pregnant woman with genotype AA and her fetus with genotype TA, the fractional concentration of fetal DNA (f) is calculated by the equation f=2xp/(p+q), where p is the occurrence of a fetal-specific allele (T allele), and q is the occurrence of an allele that occurs in both the mother and the fetus (allele A).

In another embodiment, the fractional concentration of fetal DNA in maternal plasma can be established with greater accuracy using multiple informative SNPs. When using the occurrence of alleles for multiple SNPs (total number of SNPs = n), the fractional concentration of fetal DNA (f) can be calculated using the equation _f Σ^

Σ^4/ where pi means the occurrence of a fetus-specific allele for informative SNP,;

qi - the occurrence of an allele that is common for the mother and the fetus according to the informative SNP;

and n is the total number of informative SNPs.

Using the occurrence of alleles for multiple SNPs can improve the accuracy of determining the fractional concentration of fetal DNA.

C. Fractional Concentration Without Explicit Parental Genetic Information

We now describe a method for determining the fractional concentration of fetal DNA in a maternal plasma sample that does not require prior information about the fetal and maternal genotypes. In one embodiment, informative SNPs are determined by the occurrence of different alleles at the loci of these SNPs in maternal plasma. However, the method 1000 can be used in conjunction with the determination of informative SNPs based on the embodiments described below. We first present a description of the probabilities in order to facilitate understanding of the cutoff threshold calculation used to identify informative SNPs.

In one embodiment, the probability of detecting a fetal-specific allele follows a Poisson distribution. The probability (P) of detecting a fetal-specific allele can be calculated using the following equation

P=1-exp(-fxN/2), where f is the fractional concentration of fetal DNA in the maternal plasma sample, N is the total number of molecules corresponding to the locus of this analyzed SNP, and exp() is an exponential function. In one aspect, P is considered as the expected distribution, since this distribution is not obtained by measuring the number of molecules in many samples. In other embodiments, other distributions may be used.

Assuming that the fractional concentration of fetal DNA is 5% (typical value for the first trimester of pregnancy) and subjected to analysis of 100 molecules (mother + fetus) corresponding to this SNP locus (which is equivalent to the amount contained in 50 diploid genomes), the probability of detecting a fetal-specific allele (T allele) is

1-exp(-0.05x 100/2)=0.92

The probability of detecting a fetal-specific allele increases with an increase in the fractional concentration of fetal DNA and the number of molecules analyzed per SNP locus. For example, if the concentration of fetal DNA is 10% and 100 molecules are analyzed, then the probability of finding a fetal-specific allele is 0.99.

Therefore, at the SNP locus for which the mother is homozygous, the presence in maternal plasma of an allele different from maternal may mean that this SNP is informative for calculating the fractional concentration of fetal DNA. The probability of missing any informative SNP may depend on the number of analyzed molecules. In other words, to identify informative SNPs at any desired level of confidence, the number of molecules that need to be analyzed to obtain the required accuracy can be calculated from the Poisson probability function.

Using the above analysis, in some embodiments it can be determined whether the locus is informative or not when the mother's genotype is unknown. In one embodiment, loci are determined at which two different alleles are detected in maternal plasma. For example, for the SNP locus

- 16 040939 with two possible alleles, A and T, both alleles, A and T, are detected in maternal plasma.

In FIG. 11 is a flow diagram of a method for determining 1100 that a locus is informative in accordance with embodiments of the present invention. In one embodiment, method 1100 may be applied in performing step 1020 of method 1000. In another embodiment, one step of method 1100 is to determine a cutoff based on a statistical distribution, and in another step, the cutoff is used to determine if a locus (SNP) is informative.

At 1110, a cutoff is determined to predict the occurrence of the corresponding first allele at a particular locus. In one embodiment, the cutoff predicts whether the maternal gene is homozygous and the fetal gene is heterozygous. In one embodiment, the cutoff is determined based on the statistical distribution of occurrence for various combinations of homozygosity and heterozygosity at a particular locus. For example, the allele frequency distribution can be predicted from a Poisson probability function.

In step 1120, based on the analysis of nucleic acid molecules from the parent sample (eg, in step 1010), a first allele and a second allele are detected at the locus. For example, when mapping, the analyzed locus has a set of fragments in which the first allele or the second allele was found. The first allele may correspond to one of the respective first alleles from step 1020, and the second allele may correspond to one of the respective second alleles. In one embodiment, if two different alleles are not found, then the locus is not informative.

In step 1130, the actual occurrence of the corresponding first allele at the locus is determined based on the analysis of the nucleic acid molecules. For example, you can count the results of sequencing of many nucleic acid molecules and determine how many times at this locus there is a fragment with the genotype of the first allele.

At step 1140, the locus is identified as one of the first loci based on a comparison of the actual occurrence with the cutoff. In one aspect, a cutoff is used to distinguish between three possibilities: (a) the mother is homozygous (AA) and the fetus is heterozygous (AT); (b) the mother is heterozygous (AT) and the fetus is heterozygous (AT); and (c) the mother is heterozygous (AT) and the fetus is homozygous (AA) or (TT). For illustration purposes, the examples below assume that the fetus has the AA genotype in scenario (c). However, the calculation will be the same if the fetus has the TT genotype. The informative locus corresponds to option (a).

In one embodiment, a locus is defined as one of the first loci if its actual occurrence is less than the cutoff limit. In another embodiment, a lower threshold value may be used to avoid false mapping.

Now let's describe the definition of the cutoff limit. Based on the physiologically possible fractional concentration of fetal DNA (this information is available from previous studies) and the total number of molecules corresponding to the SNP locus, it is possible to predict the distribution of allele occurrence for the three possible scenarios given above. Based on the predicted distribution, a cut-off can be determined for interpreting the observed occurrence of alleles in maternal plasma to determine if the SNP is informative (ie option (a)) or not.

The fractional concentration of fetal DNA is usually 5% to 20% at the beginning of pregnancy and 10 to 35% at the end of pregnancy (Lun et al. Micro fluidics digital PCR reveals a higher than expected fraction of fetal DNA in maternal plasma. Clin Chem 2008, 54: 1664-72). Thus, in one embodiment, the predicted distribution of allele occurrence was determined at fractional concentrations of fetal DNA of 5% and 20%.

In FIG. 12A shows the predicted distribution of occurrence for the T allele (the less common allele in scenarios (a) and (c)) for three scenarios with an assumed fractional concentration of 20% fetal DNA. In FIG. 12B shows the predicted distribution of occurrence for the T allele (the less common allele in scenarios (a) and (c)) for three scenarios with an assumed fractional concentration of fetal DNA of 5%. In both models, it is assumed that a total of 200 molecules were taken for the analysis of the SNP locus.

Using a 40-fold occurrence of the less common allele (T allele) as a cutoff, three possibilities can be statistically distinguished. In other words, if an SNP locus with two alleles is detected in maternal plasma and a total of 200 molecules are analyzed, if the frequency of a minor allele (less common allele) is less than 40, then this SNP locus can be considered informative. With a fractional concentration of fetal DNA of 5% and 20%, distinguishing informative SNPs (scenario (a)) from those SNPs for which the mother is heterozygous (scenarios (b) and (c)) will be 100% correct.

In practice, the total number of detected molecules may be different for different SNPs. For each SNP locus, a different predictive distribution curve can be drawn, taking into account the total number of detectable molecules in the maternal plasma sample covering that SNP locus. In other words, the cutoff for determining whether a SNP is informative or not may vary between different SNPs and depends on how many times that SNP locus occurs.

The following table presents the occurrence of alleles from the three SNP loci in the maternal

- 17 040939 plasma when sequencing a sample of maternal plasma. For each of the three SNPs, two different alleles were found in maternal plasma. The total number of detected cases in maternal plasma corresponding to these three SNPs differ from each other.

Locus SNP SNP designation Allele (times) Allele (times) Total number of times 1 rs3107146 A (U) G(163) 173 2 rs7522344 G(9) Т(112) 121 3 rs2272908 A (72) G(62) 134

The predicted distribution of occurrence of the less common allele at a fractional concentration of fetal DNA of 20% and various numbers of molecules corresponding to SNPs is shown in FIG. 13A, 13B and 14. The predicted distribution was made by assuming a fetal DNA concentration of 20% as it represents the upper limit of the first trimester fetal DNA concentration. The higher the concentration of fetal DNA, the more overlap is expected between the distribution curves of the minor allele for which the mother is homozygous and the major allele for which the mother is heterozygous. Thus, in order to predict informative SNPs, it would be more appropriate to calculate the cut-off for a minor allele using a higher concentration of fetal DNA.

In FIG. 13A shows the predicted distribution of occurrence of the less common allele with a total number of 173 molecules and a fetal DNA fractional concentration of 20%. In one embodiment, based on this distribution, a cut-off criterion of less than 40 for the occurrence of the less common allele is appropriate for identifying informative SNPs. Since the A allele occurs 10 times, the SNP #1 locus is considered informative for calculating the fractional concentration of fetal DNA.

In FIG. 13B shows the predicted distribution of occurrence of the less common allele at a total number of 121 molecules and a fetal DNA fractional concentration of 20%. In one embodiment, based on this distribution, a cutoff of less than 26 for the occurrence of the less common allele is appropriate for identifying informative SNPs. Since the T allele occurs 9 times, the SNP 2 locus is considered informative for calculating the fractional concentration of fetal DNA.

In FIG. 12 shows the predicted distribution of occurrence of the less common allele with a total number of 134 molecules and a fetal DNA fractional concentration of 20%. In one embodiment, based on this distribution, a cutoff of less than 25 for the occurrence of the less common allele is appropriate for identifying informative SNPs. Since the T allele occurs 62 times, SNP locus no. 3 is considered non-informative and should not be used to calculate the fractional concentration of fetal DNA.

In some embodiments, the fractional concentration of fetal DNA can be calculated using the formula f=2xp/(p+q) using the allele occurrence for SNP #1 and #2 and combining them.

The results are presented below.

SNP locus used for calculation Fractional concentration of fetal DNA 1 10x2/(10+163) = 11.6% 2 9x2/(9+112)= 14.9% 1 and 2 (10 + 9)x2 / (10 + 9 + 163 + 112) = 12.9%

D. Determination of fetal genome coverage

In addition to obtaining a fractional concentration, in embodiments, the coverage of the fetal genome by performing an analytical procedure (eg, sequencing) at step 1010 can be determined. In some embodiments, informative loci can be used to determine the coverage. For example, any of the above examples can be used. In one embodiment, loci for which the fetus is an obligate heterozygote can be used. In another embodiment, loci for which the fetus is heterozygous and the mother is homozygous can be used (eg, using method 1100).

To determine the degree of coverage, you can use fragments that, when mapped, fall on informative loci. In one embodiment, the proportion of those loci from the first set of loci in which the corresponding first allele is detected by sequencing is determined. For example, if at one locus the fetus has TA and the mother has AA, then the sequencing results should reveal the T allele if this locus has been sequenced. Thus, it is possible to calculate what proportion of the fetal genome in a biological sample has been sequenced based on this ratio. In one embodiment, the proportion of those first loci that have a fetal-specific allele is taken as the coverage of the fetal genome. In other embodiments, this proportion may be modified depending on where these loci are located. For example, it is possible to determine the degree of coverage for each chromosome. On the other hand, coverage can be considered to be less than this proportion if the first loci are not a good representative sample of the genome. On the other hand, one can imagine a range in which this proportion is one end of the range. While the high coverage rate, i.e. close to 100% means almost complete coverage of the fetal genome, most genetic diseases can be diagnosed with significantly less coverage than 100%, such as 80% or 50% or even less.

VI. Lack of preliminary information about the genome of the mother and father

In the previous sections, in some embodiments, the genetic map of the fetus (or parts of the fetal genome) was determined when the mother's haplotypes and father's genotypes were known. In other embodiments, it has been shown that the fractional concentration of fetal DNA can be determined by DNA analysis in maternal plasma without prior knowledge of maternal, paternal, or fetal genotypes. In yet other embodiments, we will now also describe a method for determining the genetic map of a fetus (or a portion of the fetal genome) by RHDO analysis without prior knowledge of maternal and paternal genotypes/haplotypes.

In one embodiment, information about control (eg, common or known) haplotypes of the population to which the parents belong is used. This information can be used to derive maternal and paternal haplotypes. Let's show the principle of this method on one example. Information on such control haplotypes can be obtained, for example, from the International HapMap Project website (hapmap.ncbi.nlm.nih.gov/).

For example, suppose there are three control haplotypes in the population (Hap A, Hap B, and Hap C, as shown in Fig. 15A). Each of these three haplotypes consists of 14 SNP loci, with two possible alleles at each locus. In this example, the father has Hap B and Hap C and the mother has Hap A and Hap B, as shown in FIG. 15b. In this example, it is assumed that the fetus inherits Hap A from the mother and Hap C from the father. Therefore, the fetus will have Hap A and Hap C as shown in FIG. 15b.

In FIG. 16 is a flow diagram of a method for determining 1600 at least a portion of the fetal genome when a set of control haplotypes is known but parental haplotypes are unknown, in accordance with embodiments of the present invention.

In step 1610, the maternal sample is analyzed to identify those SNPs for which the mother is homozygous and the fetus is heterozygous. This analysis is carried out in the same way as the determination of informative loci, as described above. Thus, in one embodiment, method 1000 and/or 1100 can be used. In other embodiments described above, maternal and paternal genomes are analyzed to obtain data for fetal genome mapping.

In FIG. 17 shows an example of determining informative loci when analyzing DNA fragments from a maternal sample. For each of the 14 loci, the occurrence of two alleles for each locus is determined. The occurrence of these alleles can be determined, for example, without limitation, real-time PCR, digital PCR, and massive parallel sequencing. For each of these loci, two different alleles must be identified in maternal plasma. Unlike those SNPs for which the mother is heterozygous, the proportion of these two alleles will be significantly different. The fetal-specific allele (the one that the fetus inherits from the father) will be much less common than the maternal allele. Informative loci 1710 are marked in FIG. 17.

At stage 1620, one or more alleles from the paternal haplotype inherited by the fetus are determined. In one embodiment, each of the loci 1710 can be used to determine the inherited paternal haplotype. For example, a paternal allele inherited by a fetus can be identified as a fetal-specific allele at the 1720 loci because the fetal-specific allele is much less common than the maternal allele in a maternal sample.

At step 1630, paternal alleles are compared with control haplotypes to determine the haplotype inherited from the father. In some embodiments, a number of possible fetal haplotypes can be deduced, each with a different probability. One or more of the most likely fetal haplotypes can then be used for further analysis or for clinical diagnosis.

In FIG. In example 18, there are three possible haplotypes in the population (Hap A, Hap B, and Hap C). In analysis of maternal plasma, four SNPs were found to be homozygous for the mother and heterozygous for the fetus, so they represent paternal alleles inherited by the fetus. The genotypes for these four SNPs matched the Hap C profile. Therefore, the fetus inherited Hap C from the father, as shown in FIG. 19. In other words, for all SNPs within the same block of haplotypes, paternal alleles inherited by the fetus can be calculated.

At 1640, loci (eg, SNPs) for which the mother is heterozygous can be determined. In one embodiment, those SNPs for which the mother is heterozygous are obtained from analysis of a maternal sample. For example, two different alleles are identified for each of these SNPs. Unlike those SNPs for which the mother is homozygous and the fetus heterozygous, where fetal-specific alleles represent only a small fraction of all alleles in maternal plasma, the occurrence of the two alleles will be the same for those SNPs for which the mother is heterozygous. Thus, by analyzing maternal plasma, it is possible to establish the complete maternal genotype across all SNP loci within a single block of haplotypes, for example, as shown in FIG. 20.

- 19 040939

At step 1650, maternal haplotypes are constructed from the mother's genotypes at step 1640 by comparing the genotypes at the loci with haplotype information in the corresponding population. In FIG. 21 shows an embodiment for determining maternal haplotypes from maternal genotypes and control haplotypes. In this example, the mother is homozygous for the G allele at the third SNP locus. Since only Hap A and Hap B meet this criterion, this means that the mother must have one of three haplotype combinations, namely Hap A/Hap A, Hap A/Hap B, or Hap B/Hap B. In addition, since the mother is heterozygous for A and C for the first SNP, it can be concluded that the mother must have the Hap A/Hap B haplotype combination. In one embodiment, more than one possibility may arise, each possibility being tested in the next step. From the above analysis, the mother's haplotypes and the haplotype inherited from the father by the fetus were determined. In FIG. 22 shows the established maternal haplotypes and the haplotype inherited from the father.

At step 1660, the maternal haplotype inherited by the fetus is determined from the haplotypes of the mother determined at step 1650 and the haplotype inherited from the father determined at step 1630. Using this information, in one embodiment, analysis is applied to determine which maternal haplotype has passed to the fetus. RHDO. RHDO analysis can be carried out according to any of the embodiments described here.

In one embodiment for RHDO analysis, those SNPs for which the mother is heterozygous can be divided into two types, alpha type and beta type (eg, as shown in FIG. 23 and described above). Alpha SNPs include those loci in which the paternal allele passed to the fetus is identical to the maternal allele in Hap A. For alpha SNPs, if the fetus inherits Hap A from the mother, alleles from Hap A will be overrepresented in maternal plasma. On the other hand, if the mother's fetus inherits Hap B, then the two maternal alleles will be equally represented in maternal plasma.

Beta SNPs include those loci in which the paternal allele passed to the fetus is identical to the maternal allele in Hap B. For beta SNPs, if the fetus inherits Hap B from the mother, alleles from Hap B will be overrepresented in maternal plasma. However, if the mother's fetus inherits Hap A, then the two maternal alleles will be equally present in maternal plasma. Possible overrepresentation of alleles from Hap A or Hap B can be determined using RHDO analysis.

In some embodiments, when applying RHDO analysis to a specific site without prior knowledge of maternal haplotypes and paternal genotypes, relatively multiple coverage of SNPs within a haplotype block may be required, for example, in one embodiment, 200 molecules corresponding to the SNP locus need to be analyzed. Such information can be obtained, for example, without limitation, real-time PCR, digital PCR, and massive parallel sequencing. In one embodiment, targeted sequencing (eg, a combination of target enrichment and massive parallel sequencing) can be used to obtain representative and unbiased quantitative information on the various alleles in the region of interest. Targeted sequencing is described in the example below. Thus, RHDO analysis can be applied to targeted maternal plasma DNA sequencing data to determine which maternal alleles/haplotypes are passed on to the fetus without prior knowledge of parental genotypes/haplotypes.

VII. Detection of de novo mutations

In some embodiments, mutations acquired by the fetus can be detected. De novo mutations are mutations that are not transmitted by the father or mother, but arise, for example, during gametogenesis in the father or in the mother, or in both. Their detection is of clinical importance, since de novo mutations play an important role in causing a number of genetic diseases, such as hemophilia A and chondrodystrophy.

In FIG. 24 is a flowchart illustrating a method for identifying 2400 de novo mutations in the genome of an unborn fetus in a pregnant woman. The fetus has a father and a pregnant mother, the father having a paternal genome with two haplotypes and the mother having a maternal genome with two haplotypes, the method comprising.

In step 2410, a plurality of nucleic acid molecules are sequenced from a biological sample taken from a pregnant woman. Note that the sample contains a mixture of maternal and fetal nucleic acids.

At step 2420, the localization of each of the sequenced nucleic acid molecules in the human genome is established. In one embodiment, sequence mapping is performed by sequencing from one end or both ends. In one aspect, mapping for localization in the human genome does not require an exact match of each of the nucleotides when detecting localization.

At 2430, for each of at least a portion of the locations, a maternal sequence and a paternal sequence at that position are determined. For example, if 100 locations are established at step 2420, then maternal and paternal genomes at those 100 positions can be determined. In one embodiment, paternal sequences are determined from a paternal sample, as opposed to using control haplotypes as described above. At the same time, it is possible

- 20 040939 identify mutations not present in the control genome. In various embodiments, maternal sequences can be obtained from a sample containing only maternal DNA, or obtained from a biological sample, for example, using the methods described here.

At step 2440, the first sequence in the plurality of nucleic acid molecules is established that is not present in the identified maternal or paternal sequences. In one embodiment, comparison of the first sequence with the identified maternal or paternal sequences requires an exact match. So, if the match is not exact, then it is considered that the first sequence is absent from the identified maternal or paternal sequences. Even minor de novo mutations can be identified, since a de novo mutation can be a change in just one nucleotide. In another embodiment, in order for a sequence to be considered a de novo mutation, a certain number of DNA fragments are required that exhibit a non-maternal and non-paternal sequence. For example, a cutoff of 3 DNA fragments can be used to determine if a sequence has or does not have a de novo mutation.

In step 2450, the first fractional concentration of the first sequence in the biological sample is determined. For example, the number of DNA fragments exhibiting a first sequence can be expressed as a proportion of all DNA fragments identified at that locus.

In step 2460, a second fractional concentration of fetal nucleic acid in the biological sample is determined using a nucleic acid molecule inherited from the father by the fetus that is present in the paternal genome but not present in the maternal genome. Such a nucleic acid molecule may contain the first allele at a locus for which the father is homozygous and the mother is also homozygous, but for a different allele, so the fetus will be an obligate heterozygote. The informative loci described above can be used to determine which nucleic acid molecule should be used to determine the second fractional concentration.

In other embodiments, the second fractional concentration is determined using other approaches, such as PCR, digital PCR, or mass spectrometry based, on the Y chromosome, on a set of genetic polymorphisms, i.e. single nucleotide polymorphisms or insertion-deletion type polymorphisms (Lun FMF et al., Clin Chem 2008, 54: 1664-1672). Another alternative is to use one or more genomic loci showing differences in DNA methylation between fetus and mother (Poon LLM et al., Clin Chem 2002, 48: 35-41; Chan KCA et al., Clin Chem 2006, 52: 2211 -2218; US Patent 6,927,028).

In one embodiment, differences in epigenetic status are reflected by differences in DNA methylation. Differences in DNA methylation may affect the 1A gene from the RAS association domain family (RASSF1A) or the holocarboxylase synthetase ligase (biotin-(propionyl-CoA-carboxylase (ATP-hydrolysing)) (HLCS) gene. the DNA methylation profile can be expressed as the proportion of all DNA fragments arising at that locus with differential methylation.

At 2470, the first sequence is classified as a de novo mutation if the first and second fractional concentrations are approximately the same. Non-maternal and non-paternal sequences resulting from errors in the analysis process, such as sequencing errors, are random events and can occur with a low probability. Therefore, multiple DNA fragments exhibiting the same non-maternal and non-paternal sequence in an amount close to the measured fractional concentration of fetal DNA in the sample most likely represent a de novo mutation present in the fetal genome and did not arise from a sequencing error. . In one embodiment, a cutoff can be used to determine that the fractional concentrations are the same. For example, if the concentrations differ from each other by a given amount, then the first sequence is classified as a de novo mutation. In various embodiments, the target value may be 5, 10 or 15%.

Examples

Example 1

To illustrate embodiments of the present invention, consider the following case. A couple visiting an obstetrics clinic for prenatal diagnosis of beta-thalassemia was recruited. The father was a carrier of a 4 bp-CTTT deletion from codons 41/42 of the human beta-globin gene. The pregnant mother was a carrier of the A^G mutation at nucleotide -28 of the human beta-globin gene. Blood samples were taken from the father and mother. The mother had a blood sample taken before chorionic villus sampling (CVS) after 12 weeks of pregnancy. Part of the CVS sample was kept for the experiment. The aim of the experiment was to map the whole genome or to determine the complete or partial genomic sequence of the fetus by massively parallel maternal plasma DNA sequencing.

1. Determination of parental genotypes

DNA was extracted from paternal and maternal buffy coats and from a CVS sample. These DNA samples were analyzed on the Affymetrix Genome-Wide Human SNP Array 6.0 system. This system contains 1.8 million genetic markers, including ~900,000 single nucleotide polymorphisms (SNPs) and more than 950,000 probes for detecting copy number variations. The absolute number and percentages of SNPs exhibiting various combinations of paternal, maternal, and fetal (CVS) genotypes are tabulated in FIG. 25A.

Although the Affymetrix system was used in this example, in practice any genotyping platform known in the art can be used. So, in addition to genotyping, DNA from the buffy coat of the father and mother can also be subjected to sequencing, both for the entire genome and for individual sections of the genome. Moreover, any source of DNA (eg, buccal cell DNA, hair follicle DNA, etc.) from the father and mother can be used to determine the genotypes of the parents.

The CVS sample was analyzed to provide a standard for comparison with the fetal genetic map from maternal plasma analysis. In addition, for this experiment, the genotype of the CVS sample can also be used to construct a maternal haplotype for RHDO analysis. However, the use of the CVS genotype to compile the maternal haplotype was used for illustration purposes only. In the clinical use of embodiments, the mother's haplotype can be derived from analysis of other family members, such as previous children, siblings, parents, or other relatives of the mother. The mother's haplotype can also be compiled from the desired chromosome regions by other methods well known to specialists, some of which are given here.

In some embodiments, the haplotype of the father of the unborn fetus is also determined. Such information can be especially useful in determining the relative dose of the haplotype for those parts of the chromosomes for which both the father and mother are heterozygous.

2. Massive parallel sequencing of maternal plasma DNA

Maternal plasma DNA was subjected to massive parallel sequencing on the Illumina Genome Analyzer platform. Sequencing of DNA molecules from plasma was performed from two ends. Each molecule was sequenced at each end by 50 bp, for a total of 100 bp. per molecule. Both ends of the sequence were aligned with the unmasked human genome (Hg18 NCBI.36 from the UCSC website http://genome.ucsc.edu) using the SOAP2 program from the Beijing Genomics Institute at Shenzhen; soap.genomics.org .cn) (Li R et al., Bioinformatics 2009, 25(15): 1966-7). In table. in fig. 25B shows alignment statistics for the first 20 flow cells. So, with 20 flow cells with a control human genome, more than 3.932 billion read segments were aligned.

3. Calculation of the fractional concentration of fetal DNA

As stated above, the fractional concentration of fetal DNA in a maternal plasma sample can be calculated from sequencing data. Thus, it is possible to analyze those SNPs for which both the father and mother are homozygous, but for different alleles. For such SNPs, the fetus will be obligate heterozygous for one paternal and one maternal allele. In one embodiment, any of the calculation methods described in section V can be used. In this example, the calculations were performed on cumulative data from various polymorphic genetic loci corresponding to the configuration of the parental genotypes (i.e. both parents are homozygous, but for different alleles) on different chromosomes. Fractional concentrations of fetal DNA calculated from SNPs located on different chromosomes are shown in the rightmost column in FIG. 26. As can be seen from the table, the fractional concentrations calculated from SNPs located on different chromosomes correlate very well with each other.

As a quality control experiment, those SNPs for which the mother is homozygous and the father is heterozygous were also examined by analyzing buffy coat samples with the Affymetrix SNP 6.0 method (middle column in Fig. 26). It can be seen that with a sufficient completeness of DNA sequencing, the fractional concentrations of fetal DNA measured in this case turned out to be very close to those measured by SNPs for which both the father and mother are homozygous, but for different alleles.

In one embodiment, if almost the same fractional concentrations of fetal DNA are observed for these two types of SNPs, then it can be concluded that the fetal genome has been almost completely covered by sequencing. In one aspect, at lower coverage, the fractional concentrations of fetal DNA measured by SNPs for which the mother is homozygous and the father is heterozygous will be higher than when measured by those SNPs for which both the father and mother are homozygous, but for different alleles. . At lower coverage, the absence of a paternally unique allele could be due to two possible factors: (i) that the fetus did not inherit the allele from the father; and/or (ii) that the fetus inherited this allele from the father, but then it was lost during sequencing because the completeness of sequencing was insufficient.

4a. Fetal genome coverage calculation

Also, as mentioned above, it is possible to determine what percentage of the fetal genome was analyzed by DNA sequencing from maternal plasma by looking at the subset of SNPs for which both father and mother are homozygous, but for different alleles. In this family, 45,900 SNPs on the Affymetrix SNP 6.0 matrix belonged to this subgroup. The coverage of the fetal genome can be calculated by analyzing plasma DNA sequencing data to find out what percentage of SNPs from this subgroup

- 22 040939 fetal allele detected by sequencing.

On the graph from Fig. 27A shows the observed percentage of SNPs in this subgroup, in which the fetal allele is observed by sequencing in the analysis of the first 20 flow cells. Thus, the fetal allele is observed in 94% of these SNPs. This degree of sequencing corresponds to more than 3.932 billion read segments with a sequence of 100 bp. On the graph from Fig. 27B shows coverage as a function of the number of segments read instead of the number of flow cells. As the throughput of various sequencing platforms increases, it is to be expected that the number of flow cells or runs used or required to obtain this many sequence reads or sequence cuts will decrease in the future.

In some embodiments, since multiple SNPs are detected in each region of the chromosome or chromosomes, the coverage of the fetal genome will be less than 94% and still provide correct mapping of the genome. For example, suppose there are 30 informative SNPs in a region of a chromosome, but the fetal allele is found in only 20 of those 30 SNPs. However, a chromosome region can be correctly identified at 20 SNPs. Thus, in one embodiment, equivalent accuracy is achieved with a coverage of less than 94%.

4b. Coverage of the alleles inherited by the fetus from the father by the genetic map

In this analysis example, those SNPs alleles for which the father is heterozygous and the mother is homozygous were examined. In this family, 131,037 SNPs on the Affymetrix SNP 6.0 matrix were in this category. The subset of such SNPs consisted of 65,875 SNPs for which the mother is homozygous and the father and fetus are heterozygous. When using 20 flow cuvettes, paternally inherited alleles were observed in 61,875 of these SNPs, representing a coverage of 93.9%. This percentage is in good agreement with the coverage data obtained in the previous paragraph. The correlations between paternally inherited allele coverage and the number of mappable sequences read and the number of sequences from flow cells are shown in FIG. 28A and FIG. 28B respectively.

To determine the specificity of this approach to identifying true paternally inherited alleles in the fetus, 65,162 (ie, 131,037 -65,875) SNPs were analyzed for which the fetus inherited the same alleles as the mother. For such SNPs, the apparent detection of alleles different from those of the mother should mean false positive results. Thus, out of these 65,162 SNPs, only 3,225 false positives (4.95%) were noted in the analysis of 20 flow cells. These false positives could result from sequencing or genotyping errors in paternal or maternal DNA, or de novo mutations in the fetus. The correlation between the proportion of false positives and the number of flow cells sequenced is shown in FIG. 29A.

The proportion of false positives can also be determined by considering the subgroup of SNPs for which both the father and mother are homozygous for the same allele. The presence of any alternative allele at a particular locus was considered a false positive. These false positives could result from sequencing or genotyping errors in paternal or maternal DNA, or de novo mutations in the fetus. There were 500,673 SNPs in this subgroup. False-positive results were reported in 48,396 SNPs (9.67%) from 20 flow cell sequencing. The correlation between the proportion of false positives and the number of flow cells sequenced is shown in FIG. 29b. The proportion of false positives in this case was higher than when assessed by the subgroup of SNPs for which both the mother and the fetus are homozygous, and the father is heterozygous. This is because in the latter subgroup of SNPs, only the presence of paternally inherited alleles in maternal plasma is considered false positive, while in the first subgroup, any allele that differs from the alleles common to the father and mother is considered a false positive.

In FIG. 30 shows the coverage of fetal-specific SNPs for various numbers of flow cells analyzed. Included in this analysis are those SNPs for which both father and mother are homozygous, but for different alleles. The x-axis represents the coverage of fetal-specific SNPs and the y-axis represents the percentage of SNPs with a given coverage.

As the number of flow cells analyzed increases, the mean coverage of fetal-specific SNPs increases. For example, with a single flow cell analyzed, the average coverage of SNPs is 0.23 times. And when analyzing 20 flow cells, the average coverage increases to 4.52 times.

5. The correctness of the genetic map inherited from the mother

In FIG. 31 shows the accuracy of Type A analysis using data from 10 flow cells. Section B describes the Type A and Type B (also referred to as alpha and beta) analysis implementations. Accuracy refers to the correct determination of the haplotype inherited from the mother. Accuracy is presented separately for each chromosome.

Using likelihood ratio =1200 in SPRT analysis (Zhou W et al., Nat Biotechnol 2001, 19:78-81; Karoui NE et al., Statist Med 2006, 25:3124-33), the accuracy ranged from 96 to 100%. As you can see, even with such a high likelihood ratio for SPRT classification, a total of 2760 segments were classified across the genome. This degree of resolution is sufficient for most problems, given that meiotic recombination occurs at a frequency of one to several times per chromosome arm per generation. In addition, it can be seen that all misclassifications could have been prevented using the interleaved approach (right in FIG. 31). As described above, the interleaved approach uses both Type A and Type B analysis.

In FIG. 32 shows the accuracy of Type B analysis using data from 10 flow cells. Using likelihood ratio = 1200 in the SPRT analysis, the accuracy ranged from 94.1% to 100%. All misclassifications could be prevented using the interleaved approach (right in FIG. 32), as noted in FIG. 31.

In FIG. 33 shows the accuracy of Type A analysis using data from 20 flow cells. Using a likelihood ratio of 1200 in the SPRT analysis and a 2-block algorithm, a total of 3780 classifications were made and only 3 (0.1%) classifications were incorrect. In FIG. 34 shows the accuracy of Type B analysis using data from 20 flow cells. Using the likelihood ratio =1200 in the SPRT analysis and the 2-block algorithm, a total of 3355 classifications were made and only 6 (0.2%) classifications were incorrect. In these examples, SPRT analysis was performed on a variety of genetic markers such as SNPs.

Example 2 Prenatal Beta Thalassemia Risk Determination.

In one embodiment, to determine the risk of fetal beta thalassemia (an autosomal recessive disease), it can be determined whether the fetus has inherited mutant alleles carried by its father and mother. In this case above, the father is a carrier of a 4 bp -CTTT deletion from codons 41/42 of the human beta globin gene. A pregnant mother is a carrier of the A^G mutation at nucleotide -28 of the human beta-globin gene.

To determine if the fetus had inherited the paternal 41/42 codon mutation, maternal plasma DNA sequencing data was searched using the first 10 flow cells for the mutation. A total of 10 signals were found with this mutation (FIG. 35A). Thus, the fetus inherited the paternal mutation. In addition, 62 signals contained the wild-type sequence at codons 41/42 (FIG. 35B). Therefore, in this region, the percentage of signals containing this mutation is 0.1389. This figure is very close to the fractional concentration of fetal DNA found in FIG. 26. In one embodiment, the risk of a fetus inheriting a paternal mutation can also be determined by the inheritance of genetic polymorphisms linked to the paternal mutation.

In one embodiment, an RHDO assay was performed to determine the risk that the fetus inherited the -28 maternal mutation. In this family, the -28 mutation is localized to haplotype IV, while the wild-type allele is localized to haplotype III. The results of the type A RHDO analysis are shown in FIG. 36 and the results of the type B RHDO analysis are shown in FIG. 37. In both types of analysis, the inheritance of haplotype III from the mother by the fetus was established. In other words, the fetus inherited the wild-type allele from the mother. The final diagnosis in the fetus is that it inherited the 41/42 codon mutation from the father and the wild-type allele from the mother. Thus, the fetus is a heterozygous carrier of betathalassemia and therefore should be clinically healthy.

Example 3 Target Enrichment and Targeted Sequencing.

As outlined in the previous sections, the accuracy of fetal DNA fractional determination and the resolution of a genetic map derived from maternal plasma DNA analysis may depend on the coverage of the loci sought. For example, we have shown that to determine with high accuracy the fractional concentration of fetal DNA without prior information about the mother's genotype, 200 molecules corresponding to the SNP locus should be enough. Allele occurrence for SNPs in maternal plasma can be obtained, for example, without limitation, by real-time PCR, digital PCR, and massive parallel sequencing.

Because massive parallel sequencing of DNA from maternal plasma can simultaneously determine the allele occurrence for millions of SNPs throughout the genome, it is an ideal platform for whole genome analysis at multiple loci. The main format of massively parallel sequencing makes it possible to cover different parts of the genome with similar completeness. However, in order to sequence a specific region with high sequencing completeness by random massive parallel sequencing, the rest of the genome (not intended for analysis) must be sequenced to the same extent. So this approach can be costly. To improve the cost-effectiveness of the massively parallel sequencing approach, it is possible to enrich the region of interest prior to sequencing. Targeted sequencing can be performed using soluble phase capture (Gnirke A et al. Solution hybrid selection with ultra-long oligonucleotides for massively parallel targeted sequencing. Nat Biotechnol 2009, 27: 182-9), microarray capture (e.g., on the NimbleGen platform ) or targeted amplification (Tewhey R et al. Microdroplet-based PCR enrichment for large-scale targeted sequencing. Nat Biotechnol 2009, 27: 1025-31).

Targeted sequencing was first applied to genetic variation within a population, such as in the study of genetic association. Therefore, its current application in genomics research is aimed at solving qualitative problems (for example, genotyping or mutation detection). However, the use of targeted DNA sequencing from maternal plasma for non-invasive

- 24 040939 prenatal diagnosis also includes quantitative considerations, the feasibility of which remains unclear. For example, the use of targeted sequencing can introduce quantitative errors in the detection of fetal and maternal DNA in maternal plasma. In addition, previous work has shown that fetal DNA is shorter than maternal DNA (Chan KCA et al. Size distributions of maternal and fetal DNA in maternal plasma. Clin Chem 2004, 50: 88-92). This size difference can also introduce quantitative errors or differences in efficiency in capturing fetal and maternal DNA in maternal plasma. In addition, there is no certainty with what efficiency such fragmented DNA molecules can be captured. In the following description, we will show that targeted sequencing can be performed by target enrichment followed by massive parallel sequencing. We will also show that target enrichment is an efficient way to assess the fractional concentration of fetal DNA compared to whole genome sequencing.

A. Fractional Concentration Determination with Target Enrichment

1. Materials and methods

4 pregnant women (M6011, M6028, M6029 and M6043) with single female fetuses were recruited. Maternal peripheral blood samples were collected in EDTA tubes before elective caesarean section in the third trimester, and placental samples were taken after elective caesarean section. After centrifugation, DNA was extracted from peripheral blood cells using the Blood Mini Kit (Qiagen). DNA from 2.4 ml of plasma was extracted using the DSP DNA Blood Mini Kit (Qiagen). Maternal genomic DNA was extracted from buffy coat and fetal genomic DNA was extracted from placental tissue. In this example, samples were taken in the third trimester for illustration purposes only. Similarly, samples from the first or second trimester can be used.

Maternal and fetal genotypes were determined using the Affymetrix Genome-Wide Human SNP Array 6.0. In each case, 5-30 ng of DNA from plasma was used to construct a DNA library using a Paired-end Sample Preparation kit (Illumina) following the manufacturer's protocol with Chromatin Immunoprecipitation Sequencing. DNA ligated to the adapter was purified directly on centrifuge columns from the QIAquick PCR Purification kit (Qiagen) without further size selection. Then, DNA ligated with the adapter was subjected to amplification for 15 cycles of PCR using standard primers, which were PCR Primer PE 1.0 and 2.0 from Illumina. DNA libraries were quantified with a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies) and passed through a 2100 Bioanalyzer using a DNA 1000 kit (Agilent) to check size distribution. For each sample, 0.6-1 μg of amplified plasma DNA library was obtained with an average size of about 290 bp. The capture library was obtained from Agilent and spanned 85% of the exons on the human chrX chromosome (Cat. No. 5190-1993). In all 4 cases in this study, 500 ng of the amplified plasma DNA library were incubated with capture probes for 24 h at 65° C. according to the manufacturer's instructions. After hybridization, captured targets were selected by attracting biotinylated probe/target hybrids with streptavidin-coated magnetic beads (Dynal DynaMag-2, Invitrogen) and purified using the MinElute PCR Purification Kit (Qiagen). Finally, DNA libraries of interest were enriched by PCR amplification for 12 cycles using Agilent's SureSelect GA PE primers. PCR products were purified using the QIAquick PCR Purification Kit (Qiagen). After that, DNA libraries obtained with or without target enrichment were subjected to random massive parallel sequencing on the Illumina Genome Analyzer Px. Sequencing used one lane on a standard flow cell to sequence one DNA library.

2. Fractional concentration of fetal DNA without target enrichment

The fractional concentration of fetal DNA can be calculated based on the occurrence of informative SNP alleles (i.e., SNPs for which the mother is homozygous and the father is heterozygous). The table below shows that for 4 cases, 120 184, LO 730, 107 362 and LO 321 informative SNPs were identified throughout the genome, while 63, 61, 69 and 65 (in the same order of listing, respectively) fall on the desired region of the X chromosome. ). Without target enrichment, fetal DNA fractional concentrations were 33.4%, 31.3%, 29.2%, and 34.4% based on all informative SNPs in the genome.

Sample Target enrichment Informative SNPs throughout the genome Shared alleles Fetal-specific alleles Fractional conc. Fetal DNA M6011 No 120 184 15 309 3064 33.4% M6028 No software 730 16 778 3 114 31.3% M6029 No 107 362 19 889 3404 29.2% M6043 No software 321 21 070 4 369 34.4%

3. Comparison of samples with and without target enrichment

In some embodiments, sequence completeness is represented by the average number of times each base has been sequenced in a particular region. In this incarnation, the fullness

-25 040939 Sequence coverage is calculated by dividing the total number of sequenced bases in the region of interest by the length of the region of interest (3.05 Mb). For the sites covered by the enrichment kit, the average sequence coverage was 0.19 times for the unenriched samples and 54.9 times for the enriched samples, which means an average of 289 times enrichment. At this completeness of sequencing, only 4.0% of fetal-specific alleles were detected in the target region before enrichment of the target (see table below). For comparison, after enrichment of the target, 95.8% of them were detected. Therefore, enrichment of the target greatly increases the rate of detection of fetal-specific alleles at a given site.

After that, we compared the fractional concentrations of fetal DNA based on the occurrence of all informative SNPs at a given site for each sample, with and without enrichment. Without target enrichment, the number of fetal-specific alleles ranged from 0 to 6 for four samples (see table below). Due to low sequence coverage, inadequate selection of fetal DNA molecules will interfere with accurate determination of the fractional concentration of fetal DNA. When the target was enriched at a given site, a much higher number of fetal-specific alleles (511-776) and co-alleles (2570-3922) were observed (see table below). The fetal DNA content was 35.4, 33.2, 26.1 and 33.0%, which is consistent with the fetal DNA content found from whole genome data in unenriched samples (see table below). These results mean that at a given site, the DNA molecules of the mother and fetus were enriched to the same extent, ____________________________

Sample Target enrichment Number of informative SNPs per site Identified spec. for fetal alleles Percentage spec. for fetal alleles Number of joint alleles Number of spec. for fetal alleles Fractional CONC. Fetal DNA M6011 No 63 6 9.5% 13 6 63.2% M6028 No 61 2 3.3% 6 2 50.0% M6029 No 69 2 2.9% eleven 2 30.8% M6043 No 65 0 0.0% 15 0 0.00% M6011 Yes 63 60 95.2% 3072 661 35.4% M6028 Yes 61 60 98.4% 2570 511 33.2% M6029 Yes 69 66 95.7% 3835 575 26.1% M6043 Yes 65 61 93.9% 3922 776 33.0%

B. Determination of the fetal genome with target enrichment

One of the applications of the RHDO method is the non-invasive prenatal diagnosis of maternally inherited genetic diseases. Using massive parallel maternal plasma sequencing without target enrichment, RHDO analysis can correctly determine which maternal haplotype was transferred to the fetus, on average, 17 SNPs, if the completeness of maternal plasma DNA sequencing is approximately 65-fold coverage of the human genome. To improve the cost-effectiveness of this approach, one can selectively target sequencing to specific regions of the genome and then apply RHDO analysis to the sequencing data. In an example, we tested this concept with targeted sequencing and RHDO analysis of chromosome X. However, targeted sequencing and RHDO analysis can be applied to all chromosomes, such as autosomes. In one embodiment, the RHDO assay as described above may be used for targeted embodiments.

5 pregnant women (PW226, PW263, PW316, PW370 and PW421) were recruited with single male fetuses. Maternal peripheral blood samples were collected in EDTA tubes prior to first trimester chorionic villus sampling (CVS). After centrifugation, DNA was extracted from peripheral blood cells using the Blood Mini Kit (Qiagen). DNA from 3.2 ml of plasma was extracted using a DSP DNA Blood Mini Kit (Qiagen). Maternal genomic DNA was extracted from buffy coat and fetal genomic DNA was extracted from chorionic villi. Samples were prepared and analyzed as described above. Each sample was then randomly sequenced using one lane on an Illumina flow cell.

In this example, the fetal genotype was used along with maternal nucleic acid sequencing information to derive maternal haplotypes for chromosome X and which haplotype is inherited from the mother. For each SNP on the X chromosome for which the mother is heterozygous (i.e., these are informative SNPs), the fetal inherited allele is determined to be derived from the maternal haplotype 1 (Hap I), while the maternal allele that is not passed on to the fetus is determined to be derived from maternal haplotype 2 (Har II). In some embodiments for clinical use, the fetal genotype may not be known in advance, and maternal haplotypes may be determined or established by methods well known to those skilled in the art, as well as by the methods described herein. The X chromosome is used for illustration purposes only. Other chromosomes, such as autosomes, can also be used in this analysis.

In all 5 cases described here, they each had one male fetus. Since the male fetus inherits only one X chromosome from the mother, but no X chromosome from the father, the maternal X chromosome that passes to the fetus will be overrepresented in the maternal plasma. Analysis

-26040939

RHDO was performed from pter to qter of chromosome X. Starting from the SNP closest to pter of chromosome X, it can be determined by SPRT analysis that an allele from Hap I or Hap II is statistically significantly overrepresented in maternal plasma. If neither of the two haplotypes is represented in an excessively statistically significant manner, then the allele occurrence can be summed for the next SNP for further analysis by the SPRT method. Additional SNPs may be pooled for analysis until the SPRT process determines that one of the haplotypes is overrepresented in a statistically significant manner. The classification process can then be restarted at the next SNP.

In FIG. 38A and B show the SPRT classification results for PW226 as an example. In this case, there were a total of 9 successful SPRT classifications for chromosome X. At each SPRT classification, alleles from Hap I were overrepresented in the maternal plasma sample, indicating that the fetus inherited Hap I from the mother. Since we identified Hap I as a haplotype containing alleles passed on to the fetus, then all of these SPRT classifications were correct.

In FIG. 39 shows the results of the RHDO analysis for these 5 cases. The number of successful SPRT classifications ranged from 1 to 9. All SPRT classifications were correct. Higher fetal DNA fractions are associated with more classifications because allelic imbalances due to the presence of fetal DNA are more readily detected at higher fetal DNA fractions. Therefore, fewer SNPs are required to obtain a successful RHDO classification. At the same time, certain regions of the chromosome can be divided into a larger number of RHDO blocks. Our results confirm that RHDO analysis can be performed on massive sequencing data obtained after target enrichment.

Our data also show that the targeted approach is a more cost-effective way to conduct RHDO analysis. Without target enrichment, when sequencing samples with similar concentrations of fetal DNA, approximately 5 flow cells (i.e. 40 sequencing lanes) are needed (Fig. 40) to achieve the same average completeness as the samples shown in Fig. 40. 39. And in our case, when enriching the target, sequencing in just one lane gives an average completeness of sequencing from 15 to 19 times for successful RHDO classification.

On the other hand, even higher sequencing coverage can be achieved at a relatively small additional cost if target enrichment is used. Higher sequencing coverage can significantly reduce the size of the genome region required for successful RHDO classification, and thereby improve analysis resolution.

Example 4. Enrichment of the target.

It has been known since 2004 that circulating fetal DNA molecules are generally shorter than maternal DNA in maternal plasma (Chan KCA et al., Clin Chem 2004, 50: 88-92; Li et al., Clin Chem 2004). However, the molecular basis of this observation remained unknown. In the present study, we read 3.931x109 segments in the studied plasma sample and used a 1 bp split. in bioinformatics analysis. The size of each sequenced DNA molecule from plasma was derived from the genomic coordinates of the ends of the segments read from both ends.

For this analysis, we focused on single nucleotide polymorphisms (SNPs) for which both father and mother are homozygous, but for different alleles. For such SNPs, the fetus is an obligate heterozygote. Alleles for each SNP inherited from the father by the fetus can be used as a fetal-specific marker. Fetal sequences (using paternally inherited fetal-specific alleles) and total sequences for the entire genome (FIG. 41) and separately for each chromosome (FIGS. 42A-C) were determined.

As it turned out, the most noticeable difference between fetal and maternal DNA in maternal plasma was a decrease in the peak at 166 bp. compared to the peak at 143 p. (Fig. 41). The most common common sequences (mainly maternal) were 166 bp long. The most notable difference in size distribution between fetal DNA and total DNA was that the fetus showed a decrease in the DNA peak at 166 bp. (Fig. 41) and the relative prominence of the peak at 143 p. The latter, apparently, corresponds to the cleavage of the connecting fragment (linker) in ~20 bp. from the nucleosome to form the nucleosome nucleus at ~146 bp. (Lewin B., in Gene IX, Jones and Bartlett, Sudbury, 2008, pp. 757-795).

Starting from about 143 p. and below, the distribution of fetal DNA and total DNA showed a periodicity of 10 bp, as in the cleavage of nucleosomes by nuclease. This indicates that DNA fragments in plasma arise under the influence of apoptotic enzymes. On the contrary, such a nucleosomal profile was not manifested when analyzing the size of the read segments corresponding to the non-histone genome of mitochondria (Fig. 41). These results provide a previously unknown molecular explanation for known size differences between fetal and maternal DNA on the Y chromosome and individual polymorphic genetic markers (Chan KCA et al., Clin Chem 2004, 50: 88-92; Li et al., Clin Chem 2004, 50: 1002-1011; US Patent Application 2005/0164241; US Patent Application 2007/0202525) and suggest that such size differences exist throughout the genome. The most probable explanation for these differences is that among the circulating fetal DNA molecules there are more molecules in which a connecting fragment of ~20 bp was cleaved from the nucleosome.

Given these observations, there are a number of ways in which a fetal DNA sample can be enriched. In one embodiment, reagents can be used that will preferentially bind to the linker moiety. Such reagents should preferentially bind to maternal DNA over fetal DNA in maternal plasma. One example of such reagents are antibodies. One of the targets of such antibodies should bind to histone H1. As is known, histone H1 binds to the linker fragment. One application of such antibodies is to enrich fetal DNA by negative selection, ie. by immunoprecipitation of maternal DNA containing the linker in maternal plasma together with histone H1. In addition, a number of H1 variants are known to exist, some of which exhibit tissue-specific variation in expression (Sancho M et al., PLoS Genet 2008, 4: el 000227). These variants could also be used to distinguish fetal DNA (predominantly from the placenta) from maternal DNA (predominantly hematopoietic) (Lui YYN et al., Clin Chem 2002, 48: 421-427). For example, a H1 histone variant that is predominantly expressed in trophoblast cells can be targeted to selectively and positively select for fetal DNA in maternal plasma. This strategy can also be applied to other histone proteins exhibiting a tissue-specific, especially trophoblast-specific, expression profile.

Considering the sharp peak at 166 p.p. in maternal DNA, another possibility for fetal DNA enrichment is to develop a negative selection system for 166±2 bp DNA fragments. For example, a system based on capillary electrophoresis or high-performance liquid chromatography would allow precise size measurement and separation of DNA molecules. Another way of negative selection is to perform it in silico during bioinformatics analysis of sequencing data.

Since other types of DNA in plasma, such as tumor DNA (Vlassov VV et al., Curr Mol Med 2010, 10: 142-165) and organ transplant DNA (Lo YMD et al., Lancet 1998, 351: 13291330), should have the same characteristics as fetal DNA in maternal plasma, then the strategies outlined in points (1) and (2) above can be used to enrich these DNA varieties.

In accordance with one embodiment, a method is provided for differential enrichment of DNA species in human plasma or serum through exposure to a nucleosome linker fragment. In one embodiment, enrichment is by removing one of the following: maternal DNA or DNA from hematopoietic cells. In another embodiment, the treatment comprises a reagent (such as an antibody or other type of protein) that will preferentially bind to the protein or nucleic acid component of the nucleosome linker fragment. In another embodiment, the reagent should selectively bind to histone H1 or other protein that binds to the nucleosome linker fragment. In another embodiment, the reagent must bind to maternal or hematologic variants of the H1 histone or other protein that binds to the nucleosome linker fragment. In one embodiment, DNA removal is carried out by immunoprecipitation or solid surface binding.

According to another embodiment, a method for differential enrichment of fetal DNA in maternal plasma or serum comprises: (a) using an antibody that binds to one or more components of a nucleosome linker fragment; (b) removing the bound fraction by immunoprecipitation or solid surface capture; and (c) recovering an unbound fraction containing fetal DNA at an increased fractional concentration.

Any components or functions of the software described in this application may be executed as a system program executable by a processor using any suitable computer language, such as Java, C++, or Perl, using, for example, standard or object-based oriented methods. A system program may be stored as a series of instructions or instructions on a computer-readable medium for storage and/or transmission, and suitable media are random access memory (RAM), read-only memory (ROM), magnetic media such as a hard disk or floppy disk, or or optical media such as compact disc (CD) or DVD (digital versatile disk), flash memory, etc. Computer-readable media can be any combination of such devices for storing or transmitting data.

Such programs may be encoded and transmitted using carrier signals adapted for transmission over wired, optical, and/or wireless networks in various protocols, including the Internet. Meanwhile, a computer-readable medium according to an embodiment of the present invention can be created with a data transmission signal encoded by such a program. The computer-readable media encoded by the control program may be packaged with a compatible device or provided separately from other devices (for example, downloaded from the Internet). Any such computer-readable media may reside on or within a single computer program product (eg, a hard drive or an entire computer system), or may reside on or within various computer program products within a system or network. The computer system may include a monitor, printer, or other suitable display to provide the user with any of the following

-

Claims (15)

here are the results. An example of a computer system is shown in Fig. 43. Shown in FIG. 43 subsystems are connected to each other via the 4375 system bus. Other subsystems are provided, such as the 4374 printer, 4378 keyboard, 4379 fixed hard drive, 4376 monitor that is connected to the 4382 display adapter, etc. Peripheral and I/O devices ( I/O) connected to the 4371 I/O controller can be connected to the computer system by any known means, such as through the 4377 serial port. For example, the 4377 serial port or the 4381 external interface can be used to connect the computer itself to a global network such as the Internet , mouse input device, or scanner. An internal connection via the system bus allows the 4373 CPU to communicate with each subsystem and control the execution of instructions from the 4372 system memory or the 4379 fixed hard drive, as well as the exchange of information between the subsystems. The system memory 4372 and/or the non-removable hard drive 4379 may implement a computer-readable medium. Any of the values given here can mean the output of data from one component to another component or to the user. The computer system may include several identical components or subsystems, for example, connected through an external interface 4381 or through an internal interface. In some embodiments, computer systems, subsystems, or machines may communicate over a network. In such cases, one computer may be considered a client and the other a server, and each may be part of the same computer system. Both client and server may include multiple systems, subsystems, or components. Specific details of certain embodiments may be combined in any suitable manner or differ from those presented and described here without departing from the essence and without going beyond the embodiments of the invention. The foregoing description of exemplary embodiments of the invention has been presented for purposes of illustration and description. It is not meant to be exhaustive or to limit the invention to the exact form described, and many modifications and variations are possible in light of the foregoing. The embodiments have been selected and described in order to best explain the principles of the invention and its practical applications, and thereby enable others skilled in the art to make the best use of the invention in various embodiments and with various modifications suitable for its particular purpose. All publications, patents and patent applications cited herein are hereby incorporated by reference in their entirety in all cases. CLAIM 1. A method for identifying de novo mutations in the genome of an unborn fetus of a pregnant woman, wherein the fetus has a father and a mother who is the specified pregnant woman, which includes obtaining the results of sequencing a plurality of nucleic acid molecules from a biological sample taken from a pregnant woman, where the biological the sample contains a mixture of maternal and fetal nucleic acids; determination of the localization of each of the sequenced nucleic acid molecules in the human genome; for each of at least part of the localizations, determining one or more maternal sequences in the maternal genome and one or more paternal sequences in that localization; identification by the computer system of a first sequence in a plurality of nucleic acid molecules that is absent from the identified maternal or paternal sequences at the first location; determining a first fractional concentration of the first sequence from the first location in the biological sample; determining the second fractional concentration of the second sequence from the second localization in the biological sample, the second sequence in the second localization is inherited by the fetus from the father, where the second sequence is present in the paternal genome in the second localization, but is absent in the maternal genome in the second localization; and classifying the first sequence as a de novo mutation if the first and second fractional concentrations are approximately the same. 2. The method of claim 1, wherein the sequencing of the nucleic acid molecules is a paired-end sequencing that provides a pair of sequences corresponding to both ends of the nucleic acid molecule, and wherein determining the location of the nucleic acid molecules in the human genome comprises mapping said both ends to a reference genome person. 3. The method of claim 1, wherein determining the location of the nucleic acid molecule in the human genome comprises mapping the nucleic acid molecule to the human genome, where mapping is not required. - 29 040939 exact sequence match. 4. The method according to claim 1, where at least one of the locations of the paternal sequence is determined from the sample from the father. 5. The method of claim 1, wherein at least one of the locations, the maternal sequence is determined from a biological sample. 6. The method according to claim 1, where the identification of the first sequence in the first localization includes comparing the sequences of nucleic acid molecules, which are identified as originating from the first localization, with one or more maternal sequences and with one or more paternal sequences in the first localization; and identifying the first sequence of the sequences as not matching any of the one or more maternal sequences and one or more paternal sequences at the first location. 7. The method according to claim 6, further comprising counting the number of nucleic acid molecules demonstrating the presence of the first sequence; and requiring that if the amount is above a certain amount for the first sequence, then it is considered a de novo mutation. 8. The method of claim 1, wherein determining the first fractional concentration comprises determining the proportion of nucleic acid molecules at the first location that exhibit the first sequence. 9. The method of claim 1, wherein the second sequence is on the Y chromosome, or is a genetic polymorphism, or a single nucleotide polymorphism, or an insertion-deletion type polymorphism. 10. The method of claim 1 wherein the mother is homozygous for the first allele at the second location and the father is homozygous for the second allele at the second location and the second allele matches the second sequence. 11. The method according to claim 1, further comprising determining that the first and second fractional concentrations are approximately the same, by calculating the difference between the first and second fractional concentrations; and comparing the difference with a threshold value. 12. A method for identifying de novo mutations in the genome of an unborn fetus of a pregnant woman, wherein the fetus has a father and a mother being said pregnant woman, where the father has the paternal genome and the mother has the maternal genome, which includes obtaining sequencing results of multiple nucleic acid molecules from a biological sample taken from a pregnant woman, the biological sample containing a mixture of maternal and fetal nucleic acids; determining the localization of each of the sequenced nucleic acid molecules in the human genome; for each of at least part of the localizations, determining one or more maternal sequences in the maternal genome and one or more paternal sequences in the paternal genome at that localization; identification by the computer system of a sequence in a plurality of nucleic acid molecules that is absent from certain maternal or paternal sequences at the first location; determining a first fractional concentration of said sequence from the first location in the biological sample; determining a second fractional concentration of fetal nucleic acids from a second location in the biological sample that exhibit a fetal-specific epigenetic status, where the epigenetic status of the fetal-derived and maternal-derived nucleic acid molecules in the biological sample is different at the second location; and classifying said sequence at the first location as de novo mutations if the first and second fractional concentrations are approximately the same. 13. The method of claim 12, wherein for at least one of the locations, the paternal sequence is determined from a sample from the paternal. 14. The method of claim 12, wherein at least one of the locations, the maternal sequence is determined from a biological sample. 15. The method according to claim 12, where the identification of the specified sequence in the first localization includes comparing the sequences of nucleic acid molecules that are identified as originating from the first localization with one or more maternal sequences and with one or more paternal sequences in the first localization; and identifying said sequence from sequences in the first localization as -