JP2015035212A - Method for finding variants from targeted sequencing panels - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for finding variants from targeted sequencing panels.SOLUTION: Provided herein is a method for identifying a sequence variant in an enriched sample. In certain embodiments, this method may comprise: (a) obtaining (i) a plurality of sequence reeds from a sample that has been enriched for a genomic region and (ii) a reference sequence for the genomic region; (b) assembling the sequence reeds to obtain a plurality of discrete sequence assemblies that correspond to potential variants; (c) determining which of the potential variants are true and which are artifacts by examining the sequence reeds that make up each of the discrete sequence assemblies; (d) optionally determining whether each of the true potential variants contains a mutation that is known to be associated with the reference sequence; and (e) outputting a report indicating whether the sample comprises a sequence variant.

Description

  The present invention relates to a method for finding mutations from a target sequencing panel.

  Comprehensive details about mutations are essential for understanding, diagnosing and treating many diseases, including cancer. A number of methods have been proposed to find mutations from sequencing data, but these usually consist of statistically assessing the presence of a mutated base relative to a reference. However, the precise determination of mutations remains a challenge in situations where mutations are found only in fragments. Such a depiction of mutations is particularly important in cancer. In order to understand the underlying cause of tumor heterogeneity, and therefore recurrence and resistance to treatment, such mutations are important not only for capturing low tumor content samples but also for capturing trace tumor subclones. is there.

  Enrichment techniques are attractive for the study of such samples due to their high homogeneity and lead depth. However, although accurate information can be obtained by experimental techniques, existing analysis methods are not suitable for detecting low-frequency mutations.

  There are many other tools, both open source and commercially available, that can call for sequence variation. Attempts to use such tools for target enrichment data often tend to be cumbersome and do not utilize all the features of the data, resulting in false calls or misjudgments and miscalls. Furthermore, as described in the literature, each method not only has its drawbacks, but the calls also do not match between the different methods. Some methods only attempt to detect low frequency mutations when a matched normal sample is provided, while others call only SNPs and insert, delete or multiple nucleotide polymorphisms (MNPs) There is also a way to not call.

  The problem is exacerbated for low frequency mutations in target sequencing with high read depth. Most methods work by looking at individual mutation sites and assessing the statistical significance of the mutation at that position. For example, if individual loci are 1000 read depths, on average, heterozygous calls are predicted to be covered by 500 reads and support mutant alleles. However, there are locations where heterozygotes really exist but are sampled only a few times. In the case of mosaic samples, mutations that are characteristic of microcomponents will have a much lower frequency. Statistically, when sampling from such a large sample space, rare events occur, making it difficult to distinguish low frequency calls from sequencing errors. The problem is further complicated by the presence of other artifacts in amplification and capture. In the presence of complex events and insertional deletions (insertion-deletions) within the genomic region, the reference sequence does not accurately represent the distribution of mutations, thereby leading to further artifacts. Many existing solutions attempt to solve this problem using multiple independent methods, but according to the latest literature, there are no solutions that can reliably call these mutations.

US Patent Application No. 20040241658 US Patent Application No. 201300120098 US Pat. No. 5,795,782 US Pat. No. 6,015,714 US Pat. No. 6,627,067 US Pat. No. 7,238,485 US Pat. No. 7,258,838 US Patent Application No. 2006003171 US Patent Application No. 20090029477 US Pat. No. 8,209,130 US Patent Application Publication No. 2011/0004413 US Patent Application Publication No. 2011/0015863 Specification US Patent Application Publication No. 2010/0063742

Hedges et al., Comparison of three targeted enrichment strategies on the SOLiD sequencing platform, PLoS One 2011 6: e18595 Shearer et al., Solution-based targeted genomic enrichment for precious DNA samples BMC Biotechnol. 2012 12: 20 Chial Proto-oncogenes to oncogenes to cancer.Nature Education 2008 1: 1 Dahl et al., Multiplex amplification enabled by selective circularization of large sets of genomic DNA fragments. Nucleic Acids Res. 2005 33: e71 Ausubel, F. M. et al., Short protocols in molecular biology, 3rd ed., 1995, John Wiley & Sons, Inc., New York Sambrook, J. et al., Molecular cloning: A laboratory manual, 2nd ed., 1989, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York Lage et al., Genome Res. 2003 13: 294-307 Zong et al., Science. 2012 338: 1622-1626 Caruccio Methods Mol. Biol. 2011 733: 241-55 Kaper et al., Proc. Natl. Acad. Sci. 2013 110: 5552-7 Marine et al., Appl. Environ. Microbiol. 2011 77: 8071-9 Margulies et al., Nature 2005 437: 376-80 Ronaghi et al., Analytical Biochemistry 1996 242: 84-9 Shendure et al., Science 2005 309: 1728-32 Imelfort et al., Brief Bioinform. 2009 10: 609-18 Fox et al., Methods Mol Biol. 2009; 553: 79-108 Appleby et al., Methods Mol Biol. 2009; 513: 19-39 Morozova et al., Genomics. 2008 92: 255-64 Soni et al., 2007 Clin. Chem. 53: 1996-2001 Myers et al., Science 2000 287: 2196-204 Batzoglou et al., Genome Research 2002 12: 177-89 Dohm et al., Genome Research 2007 17: 1697-706 Boisvert et al., Journal of Computational Biology 2010 17: 1519-33 Moreno et al., Graph-Theoretic Concepts in Computer Science 2004 3353: 168 Tarjan et al., Proc FOCS 1984 12-20 Jung et al., Systematic investigation of cancer-associated somatic point mutations in SNP databases Nature Biotechnology 2013 31: 787-789 Burmer et al., Proc. Natl. Acad. Sci. 1989 86: 2403-7 Almoguera et al., Cell 1988 53: 549-54 Tam et al., Clin. Cancer Res, 2006 12: 1647-53

  In view of the above background art, an object is to provide a method for finding a mutation from a target sequence panel.

  Provided herein are methods for identifying sequence variations in an enriched sample. In certain embodiments, the method comprises (a) (i) obtaining a plurality of sequence reads of a sample enriched in a genomic region and (ii) a reference sequence of the genomic region, (b) assembling said sequence read. To obtain a plurality of discrete sequence assemblies corresponding to the potential variation, (c) examining which sequence reads make up each of the discrete sequence assemblies, which potential variation is true Determining which are artifacts; (d) optionally determining whether each of the true potential mutations contains a mutation known to be associated with a reference sequence; And (e) outputting a report indicating whether the sample contains a sequence variation.

  A computer system is also provided that includes a memory that includes a) a database of sequences and b) an executable program for performing the method.

  A computer readable storage medium including instructions for performing the method is also provided.

  A method of identifying a mutant sequence is also provided. In certain embodiments, the method includes: a) entering sequence information into a computer system that includes a program that includes instructions for performing the method, b) executing the program, and c) from the computer system. Receiving the output.

  These and other features of the present teachings are described herein.

  Those skilled in the art will appreciate that the following drawings are for illustrative purposes only. The drawings are in no way intended to limit the scope of the present teachings.

4 is a flowchart illustrating an embodiment of the method. It is a flowchart which shows other embodiment of this method.

Definitions Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this disclosure belongs. Although any methods and materials similar or equivalent to those described herein can be used in the practice or testing of the present teachings, some representative methods and materials are now described.

  As used herein, the term “amplify” means to produce one or more copies of a target nucleic acid using the target nucleic acid as a template.

  As used herein, the term “single nucleotide polymorphism” or abbreviated “SNP” refers to a genome in which two or more alternative alleles are present in a population at a significant frequency (eg, at least 1%). Refers to a single nucleotide position in the sequence.

  The term “enriching” with respect to the genome means separating one or more genomic regions from the rest of the genome to produce a product separated from the rest of the genome. Enrichment may be performed using various methods including the methods described in Non-Patent Document 1 and Non-Patent Document 2, for example.

  The term “enriched sample” means a sample that contains genomic DNA fragments separated from the rest of the genome. The enriched fragment can be of any length depending on the fragmentation method used. In certain embodiments, fragments may be 100 bp to 1 kb in length, for example 200 bp to 500 bp in length, although fragments outside this range may be used. Depending on how fragmentation and / or enrichment is performed, the ends of the fragment molecules may be the same or different for any one enriched region.

  The term “genomic region” as used herein refers to a region of the genome, for example the genome of an animal or plant, such as a human, monkey, rat, fish or insect or plant.

“Plural” includes at least two elements. In certain instances, the plurality has at least 10, at least 100, at least 1000, at least 10,000, at least 100,000, at least 10 ⁶ , at least 10 ⁷ , at least 10 ⁸ or at least 10 ⁹ or more elements. Also good.

  As used herein, the term “sequencing” can identify at least 10 contiguous nucleotides of a polynucleotide (eg, identify at least 20, at least 50, at least 100 or at least 200 or more contiguous nucleotides). Means the method.

  The term “next generation sequencing” refers to the so-called sequencing-by-synthesis platform or ligation sequencing currently employed by Illumina, Life Technologies, and Roche. (sequencing-by-ligation) means platform. Next generation sequencing methods may also include methods based on electron detection, such as nanopore sequencing methods or ion torrent technology put to practical use by Life Technologies.

  The term “sequence read” means the output of a sequencing run. Sequence reads are represented by a single row of nucleotides. The sequence read may be accompanied by an evaluation criterion for the quality of the sequence. For example, each nucleotide in the sequence read may involve a base call reliability, ie, a determination of whether the nucleotide is G, A, T or C for that nucleotide.

  The term “sequence variation” means a nucleic acid sequence that differs from a reference sequence at at least one position. Examples of sequence variations include sequences containing SNPs and somatic mutations.

  The terms “low frequency sequence variation”, “minority species” and “minority variation” within a sample with a frequency of less than 10% (eg, less than 5% or less than 1%) relative to a non-mutated type sequence. Means an existing mutant sequence. In many cases, low frequency sequences may be represented by nucleotide substitutions or insertional deletions within the gene, and non-mutated sequences may be represented by wild-type alleles of the same gene. Infrequent sequence variations are caused, for example, by somatic mutations.

  The term “reference sequence” means a known sequence, eg, a sequence from a public or corporate database that can be compared to a candidate sequence.

  As used herein, the term “assembling” refers to a multi-step process involving alignment of sequences representing fragments of long nucleic acids. In certain cases, assembling may involve fusion of sequences to constitute the sequence of segments.

  As used herein, the term “anchor” refers to sequences present in these long sequences that can be used to align long sequences. In certain cases, the anchor may be sufficient to accurately align long sequences.

  As used herein, the term “sequence contig” means a contiguous sequence of nucleotides generated by assembling a superposed sequence.

  As used herein, the term “associated with cancer” refers to a genomic region, eg, a gene, that contains a mutation associated with a cancer phenotype. In some cases, mutations are thought to have a role as a cause of cancer.

DETAILED DESCRIPTION Before describing the various embodiments, it will be understood that the teachings of the present disclosure are not limited to the specific embodiments described, and can, of course, vary. Also, since the scope of the present teachings is limited only by the appended claims, the terminology used herein is for the purpose of describing particular embodiments only and is intended to be limiting. It will be understood that it does not.

  The section headings used in the present invention are for organizational purposes only and are not to be construed as limiting the subject matter in any way. Although the present teachings have been described in conjunction with various embodiments, it is not intended that the present teachings be limited to such embodiments. Rather, the present teachings encompass various alternatives, modifications, and equivalents, as will be appreciated by those skilled in the art.

  Where a range of values is indicated, between the upper and lower limits of the range, unless there is a clear indication in the contents, each intermediate value up to one-tenth of the lower limit unit, and the specified range It is understood that any other specified or intermediate value within is included in this disclosure.

  The citation of any document is for disclosure prior to the filing date of the application and should not be construed as an admission that the present invention ceases to be entitled to an earlier document due to the prior invention. . Also, the given release date can be different from the actual release date that needs to be independently confirmed.

  As used in this specification and the appended claims, the singular forms “a”, “an”, and “the” refer to a plural number unless the context clearly dictates otherwise. It should be noted that the target object is included. Furthermore, it is noted that the claims are written to exclude any optional element. As such, this description uses exclusive terms such as “solely”, “only”, etc., in connection with a detailed description of a claim element or the use of a “negative” limitation. It is intended to serve as an antecedent.

  It will be apparent to those skilled in the art after reading this disclosure that each individual embodiment described and shown herein is the same as that of several other embodiments, without departing from the scope or spirit of the present teachings. Each component and feature is easily separable or combined from any of the features. Any recited method can be carried out in the order of events recited or in any other order which is logically possible.

  Those skilled in the art will appreciate that the present invention is not limited in the application to configuration details, component placement, category selection, weighting, predetermined signal limits, or steps specified herein or in the drawings. Will. The invention is capable of other embodiments and of being practiced or carried out in many different ways.

  As described above, the method is obtained from a sample enriched in a specific genomic region, i.e. a sample containing fragments of genomic DNA corresponding to a specific genomic region, wherein the fragments are enriched from fragmented total genomic DNA. Alternatively, array read may be used. In some cases, the enriched genomic region comprises one or more cancers, such as breast cancer, melanoma, kidney cancer, endometrial cancer, ovarian cancer, pancreatic cancer, leukemia, colon cancer, prostate cancer, mesothelioma, glioma It may contain a gene having a mutation associated with tumor, medullobastoma, erythrocytosis, lymphoma, sarcoma or multiple myeloma (for example, see Non-Patent Document 3). Target genes include, but are not limited to, PIK3CA, NRAS, KRAS, JAK2, HRAS, FGFR3, FGFR1, EGFR, CDK4, BRAF, RET, PGDFRA, KIT and ERBB2. In certain cases, the sample is enriched with a plurality of different genomic regions (eg, several different regions, eg, at least 2, at least 5, at least 10, at least 50, at least 100, or at least 1000 or more different, overlapping A fragment of genomic DNA corresponding to a region that is not present). Each region may correspond to a gene, such as an oncogene.

  Enriched genomic regions may be enriched from the initial genomic sample using any convenient method, for example using hybridization to oligonucleotide probes or using ligation based methods. In some embodiments, to capture the region of interest, the genomic region can be 20-200 nt long in solution, such as 100-150 nt long, in one or more biotinylated oligonucleotides ( In certain cases, it may be an RNA oligonucleotide) and may be enriched by hybridization. In these embodiments, after capture, duplexes containing fragments of genomic DNA that hybridize to oligonucleotides may be separated from other fragments using, for example, streptavidin beads. In other embodiments, the region of interest may be enriched using the method described by NPL 4. In this method, a genomic sample may be denatured by fragmentation using one or more restriction enzymes. In this method, the probe library is hybridized to the target fragment. Each probe is an oligonucleotide designed to hybridize to both ends of the target DNA restriction fragment, thereby guiding the target fragment to form a circular DNA molecule. The probe also contains a method-specific sequencing motif that is incorporated during circularization. In some cases, the probe is biotinylated and the target fragment is recovered using streptavidin beads. The circular molecule is then closed by ligation, a very well defined reaction that ensures that only fully hybridized fragments are circularized. Next, the circular DNA target is amplified. Other enrichment techniques may be described in Non-Patent Document 1 and Non-Patent Document 2, for example.

  Genomic DNA may be isolated from any organism. The organism may be prokaryotic or eukaryotic. In certain cases, the organism may be a plant, such as Arabidopsis or corn, or an animal including reptiles, mammals, birds, fish and amphibians. In some cases, the initial genomic sample may be isolated from humans or rodents such as mice or rats. In exemplary embodiments, the initial genomic sample may contain genomic DNA from mammalian cells such as human, mouse, rat or monkey cells. Methods for producing genomic DNA for analysis, such as the methods described in Non-Patent Document 5 and Non-Patent Document 6, are commonly used in the art and are publicly known. The initial genomic sample contains genomic DNA or an amplified version thereof (eg, genomic DNA amplified by a whole genome amplification method using the method of Non-Patent Document 7, Non-Patent Document 8, or Published Patent Document 1). May be. Fragments can be obtained using physical methods (eg sonication, spraying or shearing), chemically, enzymatically (eg using rare-cut restriction enzymes) or transposable elements (eg Non-Patent Document 9). Non-Patent Document 10; see Non-Patent Document 11 and Patent Document 2), and may be prepared by fragmenting the genome.

  The sample may be made from cultured cells or cells of a clinical specimen, such as tissue biopsy, scrape or wash or forensic sample cells (ie, sample cells taken from a crime scene). In certain embodiments, nucleic acid samples may be obtained from biological samples such as cells, tissues, body fluids and stool. The body fluids of interest include blood, serum, plasma, saliva, mucus, sputum, cerebrospinal fluid, pleural effusion, tears, milk duct fluid, lymph fluid, sputum, cerebrospinal fluid, synovial fluid, urine, amniotic fluid and semen. However, it is not limited to these. In certain embodiments, the sample may be obtained from a subject, such as a human, and may be processed prior to use in the method. For example, the nucleic acid may be extracted from the sample before use by a known method. In certain embodiments, the genomic sample may be of a formalin fixed paraffin embedding (FFPE) sample.

  Depending on which method is performed, the initial sample (ie, prior to enrichment) may contain fragments of genomic DNA that have already been adapter ligated. In other embodiments, the fragment may be ligated to the adapter after being enriched.

  In some cases, samples may be pooled. In these embodiments, the fragment may have a molecular barcode to indicate its source. In some embodiments, the DNA to be analyzed may be from a single source (eg, a single organism, virus, tissue, cell, subject, etc.), whereas other implementations In embodiments, the nucleic acid sample may be a pool of nucleic acids extracted from multiple sources (eg, a pool of nucleic acids from multiple organisms, tissues, cells, subjects, etc.), where “multiple” Means 2 or more. As such, in certain embodiments, the sample has two or more sources, three or more sources, five or more sources, ten or more sources, 50 or more sources, 100 or more sources, 500 or more sources. Nucleic acids from sources, 1000 or more sources, 5000 or more sources up to about 10,000 sources, and about 10,000 or more sources can be included. Molecular barcodes may allow sequences from different sources to be distinguished after analysis.

  After the enriched sample is obtained, the sample is amplified and sequenced. In certain embodiments, the fragment is, for example, Illumina's reversible terminator method, Roche's pyrosequencing method (454), Life Technologies ligation sequencing (SOLiD platform) or Life Technologies' ion torrent platform. Amplified using primers suitable for use in Examples of such methods are described in the following references: Non-patent document 12; Non-patent document 13; Non-patent document 14; Non-patent document 15; Non-patent document 16; . These are incorporated by reference for a general description of the method and specific steps of the method, including the starting product, reagents and final product of each step.

  In one embodiment, the separated product may be sequenced using nanopore sequencing (eg, as described in Non-Patent Document 19 or as described by Oxford Nanopore Technologies). Nanopore sequencing is a single molecule sequencing technique in which a single molecule of DNA is sequenced directly through the nanopore. A nanopore is a small hole having a diameter of about 1 nanometer. By immersing the nanopore in a conductive fluid and applying a potential (voltage) thereto, a small current is generated by ionic conduction through the nanopore. The amount of current that flows depends on the size and shape of the nanopore. When the DNA molecule passes through the nanopore, each nucleotide of the DNA molecule blocks the nanopore to a different extent, and the magnitude of the current passing through the nanopore changes to a different extent. Thus, the change in current as the DNA molecule passes through the nanopore represents the reading of the DNA sequence. Nanopore sequencing technology is disclosed in Patent Literature 3, Patent Literature 4, Patent Literature 5, Patent Literature 6 and Patent Literature 7, and Patent Literature 8 and Patent Literature 9.

  In some embodiments, each enriched region may generate at least 100, at least 1,000, at least 10,000 to 100,000 or more sequence reads by sequencing. The sequence read length may vary greatly depending on, for example, the platform used. In some embodiments, the sequence read length may be in the range of 30-800 bases and may optionally include paired-end reads.

  A variety of different methods can be used to assemble sequence reads to obtain multiple discrete sequence assemblies, each corresponding to a potential mutation. Sequence reads are described in their basic steps in various publications such as Non-Patent Document 20, Non-Patent Document 21, Non-Patent Document 22, and Non-Patent Document 23, all of which are incorporated by reference for method disclosure. Any suitable method may be used for assembly. In some embodiments, for each enriched region, a sequence read is assembled to identify a sequence read having a nucleotide mutation (eg, substitution, insertion or deletion) at a particular position. A single pileup that can be examined can be generated. Sequence reads having nucleotide variations at specific nucleotide positions can then be reassembled as discrete sequence assemblies. In other embodiments, the sequences may be assembled with high stringency, ie, in a manner that sequence reads with the same mutation group the sequences together. In yet other embodiments, sequence reads can be assembled by aligning each read to a reference sequence, such as a reference genome. In certain cases, at least one assembled sequence obtained from a sequence read is aligned to a reference sequence.

  In some cases, and as described in more detail below, the leads are assembled using graph theory. In certain cases, assembly of sequence reads may include the creation of a directed graph such as a de Bruijn graph. For example, in a de Bruijn graph configuration of sequence reads, overlapping k-mers are collected from sequence reads including partial sequences of length k in the target region leads, and each k-mer is duplicated twice ( with k-1) -mer and assigning a vertex or node of the graph to each (k-1) -mer, and assigning an edge connecting two nodes in the graph to k-mer It's okay. Thus, each sequence read may be represented as a path through which the k-mer passes in the graph, and a potential sequence contig may be represented by combining multiple paths through which the k-mer passes in the graph. The use of the de-Bruijn graph for lead assembly is described in Patent Document 10, Patent Document 11, Patent Document 12, and Patent Document 13, which are incorporated herein by reference.

  In certain cases, the directed graph may be a directed weighted graph. In a particular aspect, the directed weighted graph is constructed using k-mers of the same length. In certain embodiments, which edges to select to construct a potential array at a node uses a cut-off value that is a function of the lead coverage of the particular node or edges connected to this node. Selected without.

  The potential array is represented by a directed weighted graph by Euler path. Thus, assembly of array leads may further involve finding Euler paths through directed weighted graphs constructed from array leads. Finding the Euler path through a directed weighted graph is a minimum de-Bruijn array (ie, every possible sub-array of length n of a given alphabet A is just a continuous character array in a language with forbidden strings. It may include finding a periodic array of A of size k that appears once. For example, see Non-Patent Document 24. In such a case, the minimal de-Bruijn array may be defined by a global subgraph of the directed weighted graph using the BEST (de Bruijn, Ehrenfest, Smith and Tutte) theorem, or by a tree (of Euler circuits in a directed graph) Provide a product formula for the number and relate the number of Euler circuits to the number of rooted spanning trees for a given vertex). The determination of the spanning tree of the directed graph may be performed by any convenient method (see, for example, Non-Patent Document 25). Representing a weighted directed graph as a de Bruijn array with prohibited words leads to an approximation of the maximum number of words possible in the graph and reflects the information entropy of the directed graph. This entropy limit is also the limit of the eigenvalue of the transition matrix of the directed graph. Since the limit of information entropy is defined by a directed graph composed of sequence reads, any potential mutant sequence that cannot be derived from a reference or other potential mutation, as there is a set of sequencing reads, It becomes unnecessary without exceeding the information entropy limit (ie, the eigenvalues of the transition matrix between the potential mutation and other mutations or references exceed the limit established above).

  In certain cases, sequence reads may be anchored to a reference sequence, which is discussed in further detail below. In some embodiments, the sequence assembly method includes demarcating regions of each of the sequence reads that are likely to be reliable for sequencing, and each assembly is unique to the reference sequence and the reference sequence. It may be anchored using a sequence.

  In this method, the sequence assembly step results in a plurality of discrete assemblies, each assembly corresponding to a potential mutation. Each potential mutation is defined by a sequence mutation found in the sequence read. Therefore, all candidate sequences of the discrete assembly have the same mutation. Any one enriched region may be represented by at least 2, at least 5, at least 10, at least 15, at least 20, at least 30, at least 50, at least 100 or more discrete assemblies. The number of sequence reads for each assembly may be highly variable. In some cases, the majority of the sequence reads may be assembled into one or two assemblies that represent the dominant mutation in the sample (the sample from which the genomic DNA was originally obtained is in an enriched region). , Depending on germline differences, eg homozygous or heterozygous for SNP). The remaining assembly may correspond to low frequency mutated sequences (eg, sequences derived from somatically mutated cells), may be derived from PCR errors, and / or may include miscalled bases. . In certain cases, these assemblies may be represented by fewer sequence reads containing mutations (eg, 10 to 1,000 or more depending on the total number of sequence reads obtained).

  In the next step of the method, the discrete assemblies are screened to determine which potential mutations are “true” (ie, correctly providing the sequence to the molecules in the sample, sequencing reactions or data Processing errors, not the result of base miscalls, and which candidate molecules are artifacts (ie, sequencing reaction or data processing errors, eg base miscalls, and the actual sequence of sample molecules Not). This step may be performed by examining the sequence reads that make up each of the discrete sequence assemblies. In some embodiments, this step may be performed by examining various parameters, including read quality, base call reliability, and alignment reliability (ie, whether the sequence has been mapped to the correct location). Good. Candidate molecules that are poorly defined (ie, candidate molecules defined by bad sequence reads, candidate molecules whose sequence variation is represented by unreliable base calls, etc.) can be canceled and the sequence aligned to other alignments Can be merged with. In certain embodiments, given a set of sequencing reads, the likelihood of each potential mutation is assigned using a hidden Markov model. In some embodiments, this step may include examining the quality of the sequence, the number of reads, the quality of the base call and its match to the reference sequence and providing a score for each potential mutation.

  Once a true potential mutation is identified, the mutation defined by the potential mutation can optionally be compared to a known mutation relative to the reference sequence. Here, the reference sequence is a public or in-company database sequence. In certain embodiments, the comparison may involve determining whether each of the true potential mutations contains a mutation known to be associated with a reference sequence. . For example, the identity of thousands of cancer-related mutations in hundreds of genes can be found in the Sanger Center's COSMIC database (see also non-patent document 26). For example, if the enriched sequence includes the sequence of the KRAS gene, the true mutation is analyzed, and then any of the sequences is 35G> A, 35G> T, 38G> A, 34G> T, 35G> C. , 34G> A, 34G> C, 37G> T, 183A> C, 37G> A, 182A> T, 183A> T, 436G> A, 37G> C, I82A> G, 34_35GG> TT, 38G> C, 181C Determine which mutation has> A, 38_39GC> AT or 38G> T. These mutations are frequently seen in leukemia, colorectal cancer (Non-patent document 27), pancreatic cancer (Non-patent document 28), and lung cancer (Non-patent document 29). Similarly, if the enriched sequence includes the sequence of the NRAS gene, the true candidate molecule is analyzed and any of the sequences is 182A> G, 181C> A, 35G> A, 182A> T, 38G> Determine which of the mutations A, 34G> A, 37G> C or 1849G> T is present in the NRAS.

  In certain embodiments, the method comprises a region where each pair of genomic regions is adjacent to (and possibly overlaps) the genomic region of interest (eg, a cancer-related gene) and the genomic region of interest. Enrichment of one or more pairs of genomic regions may be involved. In these embodiments, the pair may be enriched individually and in combination prior to amplification. Each pair of sequence reads may be analyzed together. Reading the second genomic region allows statistics to be averaged over a longer length, which yields better results. In some cases, adjacent region sequence reads can be used, for example, to adjust the results to accommodate any sampling bias.

  The method may include outputting a report indicating whether the sample contains a particular sequence variation. This report may include an indication of whether the sample contains the mutation, as well as available public information about the reference sequence and the mutation. In some cases, the report may indicate the confidence that the mutation is in the sample.

  Use the methods described above to characterize symptoms, classify symptoms, differentiate symptoms, grade symptoms, stage symptoms, diagnose symptoms or predict symptoms, or for treatment The reaction may be predicted. In certain cases, the method can be used to treat cancer symptoms or leukemia, breast cancer, prostate cancer, Alzheimer's disease, Parkinson's disease, epilepsy, amyotrophic lateral sclerosis, multiple sclerosis, stroke, autism, Other mammalian diseases may be investigated including, but not limited to, mental retardation and developmental disorders. Many nucleotide polymorphisms are associated with and are believed to be factors that cause these diseases. Knowing the type and location of nucleotide polymorphisms will greatly assist in the diagnosis, prediction and understanding of various mammalian diseases. In addition, the assay conditions described herein include, for example, infectious disease detection, viral load monitoring, viral genotyping, environmental testing, food testing, forensic medicine, epidemiology and other nucleic acid sequence detection used. It is employed in other nucleic acid detection applications including these regions.

  In some embodiments, a biological sample, such as a biopsy, may be obtained from a patient, and the sample may be analyzed using the method. In certain embodiments, the method is employed to include a biological sample comprising both a wild type copy of a genomic locus and a mutant copy of a genomic locus having a point mutation relative to the wild type copy of the genomic locus. The amount of mutated copies of the genomic locus may be identified and / or evaluated. In this example, the sample is at least 100 times (eg, at least 1,000 times, at least 5,000 times, at least 10,000 times, at least 50,000 times, or at least 100 times the mutated copy of the genomic locus. , 000 times) wild type copy of the genomic locus.

  In these embodiments, the method is employed to employ breast cancer, melanoma, renal cancer, endometrial cancer, ovarian cancer, pancreatic cancer, leukemia, colorectal cancer, prostate cancer, mesothelioma, glioma, medulloblast Carcinogenic mutations (which may be somatic mutations) that may be associated with tumors, erythrocytosis, lymphomas, sarcomas or multiple myeloma, eg, PIK3CA, NRAS, KRAS, JAK2, HRAS, FGFR3, FGFR1, EGFR, CDK4, BRAF, RET, PGDFRA, KIT or ERBB2 may be detected (see, for example, Non-Patent Document 3).

  Since point mutations at genomic loci may be directly related to cancer, the subject method alone or other clinical technique (eg, physical examination such as colonoscopy or mammogram) or Molecular techniques (eg, immunohistochemical analysis) may be employed in combination to diagnose patients with cancer or precancerous conditions (eg, adenomas, etc.). For example, the results obtained from the subject assay can include other information, such as information about the methylation status of other loci, information about rearrangements or substitutions within the same locus or at different loci, cytogenetics Combined with genetic information, information on reconstruction, gene expression information or information on telomere length, a global diagnosis of cancer or other diseases may be made.

  In one embodiment, the sample may be taken from a patient at a first location, eg, a clinical site such as a hospital or a doctor's office, and the sample may be sent to a second location, eg, a laboratory. Samples are processed at two locations and the method described above is performed to create a report. A “report” as described herein is an electronic or tangible document that contains test results that may include Ct or Cp values that indicate the presence of mutant copies of genomic loci in a sample. Contains report elements to provide. Once the report is generated, it is transferred to another location (which may be the same location as the first location), where it is a health care worker (eg, clinician, laboratory technician, or tumor) as part of a clinical diagnosis The report may be interpreted by a doctor (specialist, surgeon, pathologist, etc.).

  One embodiment of the method is described in the flowcharts of FIGS. The first flow describes the overall settings of the method, for example the overall workflow. The second flow describes the flow of the method itself. Each component of the method will now be described in detail. The method described below is an embodiment of step B3 and relates to step B4 and parts 6 and 7 of step C. In one embodiment, the method relates to identification of B3, a single nucleotide polymorphism and both insertion and deletion mutations. The flow of the present invention is described in detail in FIG.

  In step 1, design information is collected and used to annotate the target area. The design information is used in the following manner: The target area is fractionated, and the sub-area where the bait is placed is specified in the target area. Obtain and mark areas where sequencing can be assured. If desired, the specified number of bases at both ends of the region of interest can be included in the region to evaluate the off-target match of the lead and to indicate a reference anchor point for subsequent steps. . A typical reference sequence or sequences are obtained as a template. If you want to include information about any known mutations in a given region, also mark such mutations in the specified region. For efficient use of computational resources, each non-overlapping region is constructed and analyzed simultaneously (in subsequent steps) using the Java® 7 Fork-Join Framework. In this step, a “region” is just a genomic template, and data is loaded as desired and as needed. In the second step, an attempt is made to find any relevant alternative extension of the molecular sequence that can be reliably constructed in such regions. The first candidate reference sequence (s) is read from the supplied reference sequence. The method assumes that at least one molecular representation that is completely identical to the reference is obtained. If more than one such display is obtained, all are configured and evaluated as follows. Any alternative displays are then constructed. This is done by locally reassembling the target area leads. For this reassembly, we use a number of results from symbolic sequences theory, which results in optimization and quick determination of candidate molecular sequences. First, the directed weighted graph is composed of overlapping k-mers. Any candidate molecule must be represented in this graph as an Euler path (ie, through each of the edges, or in other words, edge traversal is complete). The “missed” or “unsequenced” region is considered identical to the reference and utilizes both mates of the paired end run if available. When mapping only one of the pairs with high confidence, the method looks at all unmapped leads and uses k-mer to make a local realignment implicit. Try to configure the display.

  To do this efficiently, use theoretical results. Recognizing the problem of finding a candidate solution is equivalent to finding a minimal de-Bruijn sequence in a language with forbidden strings, and the number of “words” of a particular length is used to evaluate information entropy. Note that there are limitations associated with it. This entropy limit is also the limit of the maximum eigenvalue (ie, the maximum eigenvalue is the natural logarithm of information) of the transfer matrix that specifies transitions between different k-mers. Thus, while constructing graphs representing various candidates, a count of the number of allowed words of a given length can be considered. In some cases, a count of the number of prohibited words (words that should not occur) may be considered, which gives the desired information along with the total number of possible words. Prohibited words can be easily found while constructing the graph itself. The likelihood calculation of the next step can be speeded up using the limit of the maximum eigenvalue.

  The second result used relies on the BEST theorem, the de Bruijn, Ehrenfest, Smith, and Tutte theorems. This theorem associates possible Euler paths with the number of spanning trees in the graph. Since Applicants' objective is Euler path construction, this theorem translates this problem into a problem of finding spanning trees, which is a well-known problem with a quick solution available. Spanning trees can be found using the Vishkin ’s formulation.

  Since the graph can be disequilibrium, the above results show some speedups, especially in situations where there are many duplicate matched reads or structural variations and copy number polymorphisms, although the calculations are significantly faster. Passes may be missed. In order to prevent such a corner case, the number of paths in which the incoming weight and the outgoing weight are significantly different from the average are counted. If such a path is found, the Euler path is thoroughly investigated for the partial arrangement of k-mers displayed in such a path.

  After candidate molecule representations are found, likelihoods are assigned to each using a Markov model. At this time, the lead (pair) is looked at to evaluate which candidate molecules are most likely from the given data. The leads used for this evaluation are first filtered according to a specified filtering criterion for mapping quality. The transition between candidates is represented as a transfer matrix, and the transition is optimized based on the read data of the area. During this time, the eigenvalue limits described above are used to quickly terminate any iteration that would result in a solution that does not match the limits. Output probabilities and transition probabilities are determined by standard Viterbi iterations excluding this acceleration. The specified number of highest score candidates can be examined.

  After this step, mutation calls can be made by examining the various alleles present in the candidate solution. Alleles found to be supported by bases that are too close to the lead ends ("proximity" is defined by the parameters) are filtered out. If the mutation candidate is at the end of the amplicon fragment and only one amplicon covers the gene locus, this mutation candidate is excluded by filtering. If more than one amplicon supports this locus, then such candidates are taken only if they are supported by more than one amplicon.

  Give a score for each mutation. In other words, given a set of leads {R} and a set of genotypes {G}, Applicants want to find P ({G} | {R}). To this end, Bayes' theorem is used, that is, P ({R} | {G}) and P ({G}) are obtained and combined to obtain a desired result.

  That is, given the underlying genotype, the probability of obtaining a set of leads is proportional to the probability of sampling from the set of observations of the underlying genotype, but the applicant's leads are correct Adjusted with probability. The term under the product P (b '| b) is the probability that a given substitution call at a given locus is correct. Since the quality of a base in a given lead gives the probability that Applicants excluded a read that correctly and incompletely mapped a particular base in that lead, the quality of the allele is the intermediate quality of the base and Assume the minimum of the intermediate mapping quality. If desired, base allele quality (BAQ) can be used for this assessment. If bε {G}, P (b ′ | b) is 1-q, and if other than bε {G}, P (b ′ | b) is q.

P ({G}) (possibility to see G1... Gn) by candidate molecule likelihood has already been obtained. To call a mutation at a locus, we would like to look at a site where there are two or more alleles in the candidate region and P ({G (i)} | {R (i)}) is prominent. Since the probabilities of the various candidates that are different from the reference are already known, therefore P (K> 1 | R1,..., Rn) = 1−P (K = 1 | R1,.
To get the probability of a mutation call.

  The method may be used by clinical researchers looking for a fast, accurate and easy to use analysis tool for the target enrichment panel. This software can provide an end-to-end data analysis solution, from alignment to mutation classification, reducing the time to results from days to hours. This method has a much lower detection failure rate in mutation calls without affecting the misjudgment rate for the majority of test samples, and this method allows low frequency alleles even in complex cases involving multiple alleles. It is possible to detect mutations that have a gene, and at the same time, it does not significantly increase the misjudgment rate, and when detecting low-frequency mutations, the efficiency and speed do not decrease significantly. is there.

  The method described above can be implemented on a computer. In certain embodiments, a general purpose computer can be configured into a functional structure for the methods and programs disclosed herein. The hardware architecture of such computers is well known to those skilled in the art and includes one or more processors (CPU), random access memory (RAM), read only memory (ROM), internal or external data storage media (eg, hard disk drive). ) Including hardware components. The computer system may also include one or more graphic boards for processing and outputting graphic information on the display means. The above components can be appropriately interconnected by a bus in the computer. The computer further includes a suitable interface for communicating with general purpose external components such as a monitor, keyboard, mouse, network and the like. In some embodiments, to increase processing power for the present methods and programs, the computer can be parallel or be part of a network configured for parallel or distributed computation. it can. In some embodiments, the program code read from the storage medium can be written into an expansion board embedded in the computer or a memory provided in an expansion unit connected to the computer. Alternatively, a CPU or the like provided in the expansion unit can actually perform some or all of the calculations according to the instructions of the program code in order to achieve the following functions. In other embodiments, the method can be implemented using a cloud computing system. In these embodiments, data files and programming can be exported to a cloud computer, which executes the program and returns output to the user.

  The system, in certain embodiments, includes: a) a central processing unit; b) a main non-volatile, where the storage drive for storing software and data can be controlled by a disk controller; C) a system memory for storing system control programs, data and application programs, including programs and data loaded from non-volatile storage drives, eg high speed random access memory (RAM) (in system memory) Can include read-only memory (ROM)), d) a user interface including one or more input and output devices such as a mouse, keypad and display, e) any wired or wireless communication network, eg For connecting to a printer, comprising a computer including any network interface card, and an internal bus for interconnecting the above elements of f) system.

  The memory of a computer system is any device that can store information for retrieval by a processor and may include magnetic or optical devices or solid state memory devices (such as volatile or non-volatile RAM). A memory or memory unit may have two or more physical memory devices of the same or different types (eg, a memory may be a plurality of memory devices such as a plurality of drives, cards, or a plurality of solid state memory devices or these Can have several combinations). With respect to computer readable media, “permanent memory” means permanent memory. Permanent memory does not disappear when power to the computer or processor is stopped. Computer hard drive ROM (ie, ROM not used as virtual memory), CD-ROM, floppy disk and DVD are all examples of permanent memory. Random access memory (RAM) is an example of non-permanent (ie, volatile) memory. Files in permanent memory can be editable and rewritable.

  Computer operations are controlled primarily by the operating system, which is performed by a central processing unit. The operating system can be stored in system memory. In some embodiments, the operating system includes a file system. In addition to the operating system, one possible implementation of system memory includes various programming and data files for performing the methods described below. In certain cases, programming can include a program that can be composed of various modules and user interface modules that allow the user to manually select or change the inputs to the program or the parameters used in the program. The data file can contain various inputs for the program.

  In certain embodiments, instructions according to the methods described herein can be encoded on a computer-readable medium in the form of “programming”. The term “computer-readable medium” as used herein refers to any storage or transmission medium that participates in providing instructions and / or data to a computer for execution and / or processing. Examples of storage media include floppy disks, hard disks, optical disks, magneto-optical disks, CD-ROMs, CD-Rs, magnetic tapes, non-volatiles, whether such devices are internal or external to the computer. Volatile memory card, ROM, DVD-ROM, Blu-ray disc, solid state disc, and network attached storage (NAS). A file containing information can be “saved” on a computer readable medium, where “save” means storing the information so that the information is accessible and searchable at a later date by the computer.

  The computer-implemented methods described herein can be performed using a program that can be written in any number of one or more computer programming languages. Such languages include, for example, Java (Sun Microsystems, Santa Clara, CA), Visual Basic (Microsoft, Redmond, WA) and C ++ (AT & T, Bedminster, NJ) and any Including many multi-languages.

  In any embodiment, data can be transferred to a “remote location”, where “remote location” means a location other than where the program is executed. For example, remote locations can be other locations in the same city (eg, offices, laboratories, etc.), other locations in different cities, other locations in other states, other locations in different countries, etc. Thus, when one item is shown to be “remote” to another item, the two items are in the same room but are separated, or at least in different rooms or different buildings, and at least 1 mile, 10 Means miles or at least 100 miles away. “Communication” information means the transmission of data representing that information as an electrical signal on a suitable communication channel (eg, a private or public network). “Transfer” of an item means any means of moving the item from one place to the next, either by physically transporting the item or otherwise (if possible) This includes, at least in the case of data, physically transporting the medium that holds the data or communicating the data. Examples of communication media include wireless or infrared transmission lines and network connections to other computers or network devices, and the Internet, or include information stored in e-mail transmissions and websites.

  Some embodiments include a handheld computer on a local computer, on a single computer, on a computer network, or on a network of computer networks, eg, on a cloud of networks, etc. Embodiments. Preferred embodiments include implementations in computer program (s) that perform one or more of the steps described herein. Such a computer program performs one or more of the steps described herein. Preferred embodiments of the present invention are various data structures, categories, and modifications that are encoded on computer readable medium (s) and that can be transmitted over communication network (s) as described in the present invention. Includes children.

  Software, web, internet, cloud or other storage and computer network implementations of the present invention are standard programming to achieve searching, modifying, associating, comparing, determining, signaling, scoring, monitoring or ranking various databases Could be achieved with technology.

  All publications and patent applications cited herein are hereby incorporated by reference as if each individual publication or patent application was clearly and individually indicated to be incorporated by reference. The The citation of any document is for disclosure prior to the filing date of the application and should not be construed as an admission that the present invention ceases to be entitled to an earlier document due to the prior invention. .

This application claims the benefit of US Provisional Patent Application No. 61 / 859,625 (filed July 29, 2013), which is hereby incorporated by reference in its entirety. ing.

Claims (19)

A method for identifying a sequence variation comprising:
(A) (i) obtaining a plurality of sequence reads of a sample enriched in a genomic region and (ii) a reference sequence for the genomic region;
(B) combining the sequence reads to obtain a plurality of discrete sequence assemblies, each corresponding to a potential mutation;
(C) determining which potential mutations are true and which are artifacts by examining the sequence reads comprising each of the discrete sequence assemblies;
(D) optionally, determining whether each of the true potential mutations contains a mutation known to be associated with the reference sequence; and (e) whether the sample contains a sequence mutation. Outputting a report indicating whether or not.   The method of claim 1, wherein the genomic region is associated with cancer.   2. The genomic region comprises at least a portion of at least one of the following genes: PlK3CA, NRAS, KRAS, JAK2, HRAS, FGFR3, FGFR1, EGFR, CDK4, BRAF, RET, FGDFRA, KIT and ERBB2. the method of.   The method of claim 1, wherein the sequence variant is a low frequency sequence variation corresponding to a somatic mutation.   The method of claim 1, wherein the genomic region is a region of the human genome.   2. The method of claim 1, wherein the enriched genomic region is enriched from total DNA obtained from a clinical specimen.   The method of claim 6, wherein the clinical specimen is a biopsy.   The method of claim 1, wherein the report provides an indication of whether the specimen contains a mutation and public information available about the reference sequence.   The method of claim 1, wherein the assembling comprises fractionating each of the regions of the sequence read where the sequence is likely to be reliable.   The method of claim 1, wherein the assembling uses graph theory.   The method of claim 10, wherein the assembling is performed using a minimal de-Bruijn array.   The method of claim 10, wherein the assembling is performed using a BEST theorem.   The determination of claim 1, wherein the determining comprises examining sequence quality, number of reads, base call quality and its match to the reference sequence and providing a score for each of the potential mutations. the method of.   The method of claim 1, wherein the reference sequence is known in the art and is annotated to identify mutations for which a sequencing read is appropriate.   The method of claim 1, wherein the assembling uses a sequence from the reference sequence and a sequence unique to the reference sequence to anchor the assembly.   The method of claim 1, wherein the method provides a probability of a mutant call. A computer system including a memory,
(A) a sequence read database of samples enriched in genomic regions;
A computer system comprising (b) a reference sequence of the genomic region, and (c) a program executable to perform the method of claim 1   A computer readable storage medium comprising instructions for performing the method of claim 1. A method for identifying a mutant sequence, comprising:
a) inputting sequence information into a computer system comprising a program comprising instructions for performing the method of claim 1;
b) executing the program; and c) receiving output from the computer system.