WO2016205767A1 - Assemblage de graphes de chaînes pour génomes polyploïdes - Google Patents

Assemblage de graphes de chaînes pour génomes polyploïdes Download PDF

Info

Publication number
WO2016205767A1
WO2016205767A1 PCT/US2016/038264 US2016038264W WO2016205767A1 WO 2016205767 A1 WO2016205767 A1 WO 2016205767A1 US 2016038264 W US2016038264 W US 2016038264W WO 2016205767 A1 WO2016205767 A1 WO 2016205767A1
Authority
WO
WIPO (PCT)
Prior art keywords
string
graph
paths
contigs
sequence
Prior art date
Application number
PCT/US2016/038264
Other languages
English (en)
Inventor
Chen-Shan Chin
Original Assignee
Pacific Biosciences Of California, Inc
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Priority claimed from US14/743,713 external-priority patent/US20150286775A1/en
Application filed by Pacific Biosciences Of California, Inc filed Critical Pacific Biosciences Of California, Inc
Publication of WO2016205767A1 publication Critical patent/WO2016205767A1/fr

Links

Classifications

    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G16B20/10—Ploidy or copy number detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations
    • G16B20/20—Allele or variant detection, e.g. single nucleotide polymorphism [SNP] detection
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B30/00—ICT specially adapted for sequence analysis involving nucleotides or amino acids
    • G16B30/20—Sequence assembly
    • G—PHYSICS
    • G16—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR SPECIFIC APPLICATION FIELDS
    • G16B—BIOINFORMATICS, i.e. INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR GENETIC OR PROTEIN-RELATED DATA PROCESSING IN COMPUTATIONAL MOLECULAR BIOLOGY
    • G16B20/00—ICT specially adapted for functional genomics or proteomics, e.g. genotype-phenotype associations

Definitions

  • a chromosome can have loci that differ in sequence from the homologous chromosome. When these loci are sequenced, the base calls will differ between the homologous chromosomes. It is important to be able to determine whether basecalls that differ between homologous chromosomes are true variations between the homologues, or are merely sequencing errors.
  • a viral population in an individual can have many variations between individual viral genomes in the population, especially in highly mutable viruses such as HIV.
  • Being able to identify different sequencing reads that have different origins e.g., different chromosome or genome origins
  • the reads can simply be compared to one another with simple string matching algorithms. Any difference between the reads is indicative of a true variant, and therefore, a different origin.
  • any real-world raw sequencing data is likely to contain errors, so a simple string matching algorithmic approach will not be sufficient.
  • a string graph is a data structure that can be used to model a genome, e.g., to aid in assembling the genome from sequencing data. Modeling a genome with a string graph has generally advantages over modeling with an overlap graph or a de Brujin graph. For example, both correction of sequence and/or consensus errors and annotation of heterogeneous regions may be improved. For further details on string graph construction, see Fragment assembly string graph, Myers, E. W. (2005) Bioinformatics 21 (iss. suppl. 2):ii79-N85), of which is incorporated herein by reference.
  • a vertex (also called a node) is a beginning and/or end of a sequence fragment, and an edge is the sequence fragment between two vertices.
  • the core of the string graph algorithm is to convert each "proper overlap" (where only a portion of each of two reads overlaps the other read, i.e., the first read extends beyond the overlap at the 3' and the second read extends beyond the overlap at the 5' end) between two fragments into a string graph structure. This process comprises identifying vertices that are at the edges of an overlapping region and extending the edges to the non- overlapped parts of the overlapping fragments.
  • the edge is labeled depending on the direction of the sequence and redundant edges are removed by transitive reduction to yield the string graph.
  • this de-tangling will generate two complementary contigs, one for the forward strand and one for the reverse strand, which can be further reduced to a single contig that represents the genome assembly.
  • Additional features observed in string graph structures include branching, knots, and bubbles.
  • Branching or branch points are typically caused when the reads contain some repetitive sequence, e.g. due to repeat regions in the genome.
  • Knots, where many edges connect to the same node can be caused by many reads that contain the same repeat in the genome.
  • a simple "best overlapping logic" is typically used to "de-tangle" simple knots.
  • Simple bubbles are generally observed where there are local structural variations, and are usually easy to resolve. However, simple bubbles can also be caused by errors in the original sequence reads and/or in the consensus determination performed during the pre-assembly of the reads.
  • the overlap identification step fails to detect a proper overlap, a bubble will be rendered in the string graph.
  • the invention is generally directed to processes for analyzing sequence data from mixed populations of nucleic acids, for assigning each sequence read to a particular origin, and for ultimately identifying one or more consensus sequences of one or more biomolecular target sequences from the sequence information.
  • the methods provided herein are applicable not only to sequence data having few errors, but also to sequence data having relatively high rates of insertions, deletions, and/or mismatch errors. Consequently, the invention is also directed to systems that carry out these processes.
  • the methods are beneficial for sequencing polyploid organisms wherein the sequence reads are assigned to a specific homolog.
  • the invention provides methods for string graph assembly of polyploid genomes, the method performed by at least one software component executing on at least one processor.
  • such methods comprise several steps including receiving a string graph generated from long sequence reads of at least .5 kb in length through 1 , 2, 3, 4, 5, 7, or 10 kb in length; identifying unitigs in the string graph and generating a unitig graph; and identifying string bundles in the unitig graph by: determining one or more end-to-end paths through each component of the unitig graph that are then used to define corresponding string bundles and a primary contig for each of the string bundles; and determining associated contigs that contain structural variations compared to the primary contig.
  • the output from the method may include the primary contigs, the associated contigs, and the string graph.
  • receiving the string graph may further include conducting a pre-assembly step that includes aligning nucleic acid sequence reads to each other; detecting overlaps between the aligned reads; determining consensus sequences from the aligned reads, and constructing the string graph from the consensus sequences.
  • the aligning operation may comprise choosing a best-match sequence read from the nucleic acid sequence reads as a seed sequence; aligning the nucleic acid sequence reads to the seed sequence to generate a set of sequence alignments; and using the set of sequence alignments to construct a sequence alignment graph (e.g., string/unitig graph).
  • a plurality of consensus sequences are generated, each corresponding to a different nucleic acid in the original sample.
  • the exemplary embodiment may further include performing additional steps in diploid assembly based on the primary and associated contigs. Aspects of this embodiment may include responsive to determining the primary contigs and the associated contigs, analyzing an allelic constitution of the sequence reads to determine whether a single sequence read contains more than one variant positions, including multinucleotide structural variations and/or single nucleotide polymorphisms (SNPs); responsive to determining that the single read contains more than one of the variant positions and therefore that the alleles at those loci are linked, identifying the loci as originating from a single original nucleic acid molecule; and determining which version of each variant position originates with which nucleic acid molecule, thereby determining a final consensus sequence for the nucleic acid molecules.
  • SNPs single nucleotide polymorphisms
  • the invention provides methods for string graph assembly of polyploid genomes, the method performed by at least one software component executing on at least one processor.
  • such methods comprise several steps including receiving a string graph generated from long sequence reads of at least .5 kb in length through 1 , 2, 3, 4, 5, 7, or 10 kb in length; identifying bubble regions as compound paths in the string graph by: simplifying the string graph to a first graph with simple paths in which edges in a path without any branching node are represented with a single edge; finding nodes having multiple out-edges in the first graph, and for each of these nodes, a initiating search to find a local bundle of edges; for compound paths that are overlapped with others and for nested compound paths, selecting the longest compound path and ignoring the smaller compound path; and generating a new graph in which each of the compound paths identified in the first graph, are replaced by a single compound edge, and each of the simple paths in the first graph connecting the compound paths are replaced with a simple edge; and
  • Figure 1 is a diagram illustrating one embodiment of a computer system for implementing a process for using a string graph to assemble a diploid or polyploid genome.
  • Figure 2 is a flow diagram illustrating a process for string graph assembly of polyploid genomes according to an exemplary embodiment.
  • Figures 3A and 3B are diagrams illustrating embodiments of methods for creating a string graph from overlaps between aligned sequences and an algorithm for transitive reduction.
  • Figure 4 is a diagram graph illustrating an exemplary string graph generated for a double-stranded haploid sample, e.g., E. coli genome.
  • Figure 5 is a diagram showing that additional features observed in string graph structures may include areas of entanglement such as branching, knots, and bubbles.
  • Figure 6 is a diagram showing results of applying the best overlapping rule on the E. coli string graph.
  • Figure 7 is a diagram graphically illustrating identifying unitigs from the non- branching parts of the string graph to generate a unitig graph.
  • Figure 8A is a diagram graphically illustrating that a string graph that may have a quasi linear structure and bubbles.
  • Figure 8B is a diagram illustrating that simple bubbles can also be caused by errors in the original sequence reads and/or in the consensus determination performed during the pre-assembly of the reads.
  • Figure 9 is a diagram illustrating one challenge of diploid assembly is to determine the genetic sequence underlying complex structures in a string graph where the same structure in the string graph can be caused by repeats or the presence of homologous sequences.
  • Figure 10 is a diagram graphically illustrating exemplary large- and small-scale topological features of a unitig graph.
  • Figure 1 1 is a diagram graph illustrating the processing of string bundles according to a first embodiment.
  • Figure 12 is a diagram illustrating a process for determining whether a junction at a vertex in a unitig graph belongs to a string bundle or a branching path.
  • Figure 13 is a diagram graph illustrating the processing of string bundles according to a second embodiment.
  • Figure 14 is a diagram graphically illustrating construction of a final consensus sequence based on the primary contigs and the associated contigs.
  • FIG. 1 is a diagram illustrating one embodiment of a computer system for implementing a process for using a string graph to assemble a diploid or polyploid genome.
  • the invention may be embodied in whole or in part as software recorded on fixed media.
  • the computer 100 may be any electronic device having at least one processor 102 (e.g., CPU and the like), a memory 103, input/output (I/O) 104, and a data repository 106.
  • the CPU 100, the memory 1 02, the I/O 104 and the data repository 106 may be connected via a system bus or buses, or alternatively using any type of communication connection.
  • the computer 100 may also include a network interface for wired and/or wireless communication.
  • computer 100 may comprise a personal computer (e.g., desktop, laptop, tablet etc.), a server, a client computer, or wearable device.
  • computer 100 may comprise any type of information appliance for interacting with a remote data application, and could include such devices as an internet-enabled television, cell phone, and the like.
  • the processor 102 controls operation of the computer 100 and may read information (e.g., instructions and/or data) from the memory 103 and/or a data repository 106 and execute the instructions accordingly to implement the exemplary embodiments.
  • information e.g., instructions and/or data
  • the term processor 102 is intended to include one processor, multiple processors, or one or more processors with multiple cores.
  • the I/O 1 04 may include any type of input devices such as a keyboard, a mouse, a microphone, etc., and any type of output devices such as a monitor and a printer, for example.
  • the output devices may be coupled to a local client computer.
  • the memory 103 may comprise any type of static or dynamic memory, including flash memory, DRAM, SRAM, and the like.
  • the memory 103 may store programs and data including a sequence aligner/overlapper 1 10, a string graph generator 1 12 and a diploid contig generator 1 14. These components are used in the process of sequence assembly as described herein, and are generally referred to collectively as the "assembler.”
  • the data repository 1 06 may store several databases including one or more databases that store nucleic acid sequence reads (hereinafter, “sequence reads”) 1 16, aligned sequences 1 17, a string graph 1 18, a unitig graph 120, primary contigs 1 22, associated contigs 1 24, final assembly graph 126, and haplotype data 128.
  • sequence reads nucleic acid sequence reads
  • the data repository 106 may reside within the computer 1 00. In another embodiment, the data repository 1 06 may be connected to the computer 100 via a network port or external drive.
  • the data repository 106 may comprise a separate server or any type of memory storage device (e.g., a disk-type optical or magnetic media, solid state dynamic or static memory, and the like).
  • the data repository 106 may optionally comprise multiple auxiliary memory devices, e.g., for separate storage of input sequences (e.g., sequence reads, reference sequences, etc.), sequence information, results of string graph generation (e.g., edges and nodes in a string graph, overlaps and branch points in a string graph), results of transitive reduction, and/or other information.
  • Computer 100 can thereafter use that information to direct server or client logic, as understood in the art, to embody aspects of the invention.
  • an operator may interact with the computer 100 via a user interface presented on a display screen (not shown) to specify the sequence reads 1 16 and other parameters required by the various software programs.
  • the programs in the memory 103 including the sequence aligner/overlapper 1 10, the string graph generator 1 1 0, and the diploid contig generator 1 14, are executed by the processor 102 to implement the methods of the present invention.
  • sequence aligner/overlapper 1 10 reads the selected sequence reads 1 16 from the data repository 106 and performs sequence alignment on the selected sequence reads 1 16 to identify regions of similarity that may be a consequence of structural or functional or other relationships between the sequence reads 1 16.
  • Sequence reads 1 16 are generally high accuracy reads, e.g., at least about 98% or 99% accurate, and may be raw reads from a sequencing technology that provides such high quality reads, or may be pre-assembled, high-quality reads constructed from sequencing read data of a lower quality, as described elsewhere herein.
  • Aligned sequences 1 17 are generated by the sequence aligner/overlaper 1 10 during the sequence alignment.
  • the sequence aligner/overlaper 1 1 0 is implemented in C, C++, Java, C#, F#, Python, Perl, Haskell, Scala, Lisp, a Python/C hybrid, and others known in the art.
  • the string graph generator 1 1 2 receives the resulting aligned sequences 1 1 7 and may generate the string graph 1 1 8 as well as the unitig graph 120 from the aligned sequences 1 17.
  • the diploid contig generator 1 14 analyzes the string graph 1 18 and the unitig graph 120 and determines the primary contigs 122 and associated contigs 124 in accordance with exemplary embodiments, as explained further below.
  • the progress and/or result of this processing may be saved to the memory 103 and the data repository 106 and/or output through the I/O 104 for display on a display device and/or saved to an additional storage device (e.g., CD, DVD, Blu-ray, flash memory card, etc.), or printed.
  • the result of the processing may include the primary contigs 122, the associated contigs 124, and optionally the string graph 1 1 8 and haplotype data 128, which be stored or displayed in whole or in part, as determined by the practitioner.
  • the results may further comprise quality information, technology information (e.g., peak characteristics, expected error rates), alternate (e.g., second or third best) final assembly graph 126, confidence metrics, and the like.
  • One of the main challenges in assembling diploid or polyploid genomes is that it is often difficult to distinguish between homologous sequences on different chromosomes, e.g., to identify individual haplotypes for the homologous chromosomes, or to analyze the size of a repetitive region, e.g., to determine the number of repeats in each homolog.
  • Standard assembly algorithms assume the sequence reads all come from the same original nucleic acid molecule (e.g., chromosome). Conventional assembly algorithms often create a graph structure.
  • the conventional assembly algorithms typically break resulting contigs at a junction where there is a fork in the assembly graph, e.g., unitig graph, overlap graph, string graph, De Bruijn graph, and the like, e.g., where the fork is due to sequence differences between the homologs.
  • a fork in the assembly graph
  • Algorithmica 13 (1 -2):7-51 e.g., Algorithmica 13 (1 -2):7-51
  • Myers E. W. (2005) Bioinformatics 21 (iss. suppl. 2):ii79-N85), both of which are incorporated herein by reference in their entireties for all purposes.
  • the exemplary embodiments are generally directed to powerful and flexible methods and systems for string graph assembly of polyploid genomes using long reads that generate long contigs comprising structural differences that distinguish between homologous sequences from multiple different nucleic acid molecules, repetitive sequences within a single nucleic acid molecule, and repetitive sequences within homologous sequences.
  • Figure 2 is a flow diagram illustrating certain aspects of a process for string graph assembly of polyploid genomes according to an exemplary embodiment.
  • the process may be performed by the diploid contig generator 1 14 executing on the processor 102.
  • the process may begin by the receiving a string graph generated from sequence reads of at least 0.5 kb, more preferably of at least 1 kb in length (block 200). Unitigs are identified in the string graph and a unitig graph is generated (block 202). String bundles are identified in the unitig graph (block 204).
  • a string bundle may comprise a set of non-branching edges that form compound and simple paths that contain sequences from both haplotypes. Block 204 includes two sub-steps.
  • the string bundles are identified by determining one or more end-to-end paths through each component of the unitig graph that are then used to define corresponding string bundles and a primary contig for each of the string bundles (block 204A).
  • a primary contig 1 102 is defined as a single path that extends the length of the string bundle that may be the same or different as the end-to-end path used to find the string bundle.
  • the primary contig may represent a single template molecule, or may represent more than one homologous template molecule, at least in regions where the homologs do not differ in sequence.
  • associated contigs that contain structural variations compared to the primary contig are then determined (block 204B).
  • associated contigs are paths in parallel to the primary contig in bubble regions of the string bundle.
  • associated contigs often represent regions in which the homologous templates comprise sequence differences, e.g., SNPs, structural variations, mutations, etc.
  • the process may further include identifying candidate break points in the primary contigs; and breaking the corresponding primary contigs at the break points.
  • the string graph 1 18 may be generated by the string graph generator 1 12, which in turn, uses as input the aligned sequences 1 17 generated by the sequence aligner/overlaper 1 10 from the sequence reads 1 1 6.
  • the string graph 1 18 may be generated on another computer or received from a third party for subsequent input to the diploid contig generator 1 14.
  • the sequence reads 1 1 6 used as input to generate the string graph 1 18 are considered long sequencing reads, ranging in length from about 0.5 to 1 , 2, 3, 5, 10, 15, or 20 kb.
  • these long sequencing reads are generated using a single polymerase enzyme polymerizing a nascent strand complementary to a single template molecule.
  • the long sequencing reads may be generated using Pacific Biosciences' single- molecule, real-time (SMRT®) sequencing technology.
  • SMRT® real-time
  • flanking sequence in a read comprising a repeat region allows the practitioner to accurately map these sequence variants within the repetitive region. This is difficult or impossible with short sequence reads, especially where the variants occur far from the flanking sequence.
  • the sequence reads 1 16 may be generated using a single- molecule sequencing technology such that each read is derived from sequencing of a single template molecule.
  • Single-molecule sequencing methods are known in the art, and preferred methods are provided in U.S. Patent Nos. 7,315,019, 7,476,503, 7,056,661 , 8,153,375, and 8,143,030; U.S.S.N. 12/635,61 8, filed December 10, 2009; and U.S.S.N. 12/767,673, filed April 26, 2010, all of which are incorporated herein by reference in their entirety for all purposes.
  • the technology used comprises a zero-mode waveguide (ZMW).
  • sequence reads from various kinds of biomolecules may be analyzed by the methods presented herein, e.g., polynucleotides and polypeptides.
  • the biomolecule may be naturally-occurring or synthetic, and may comprise chemically and/or naturally modified units, e.g., acetylated amino acids, methylated nucleotides, etc. Methods for detecting such modified units are provided, e.g., in U.S.S.N. 12/635,618, filed December
  • the biomolecule is a nucleic acid, such as DNA, RNA, cDNA, or derivatives thereof.
  • the biomolecule is a genomic DNA molecule.
  • the biomolecule may be derived from any living or once living organism, including but not limited to prokaryote, eukaryote, plant, animal, and virus, as well as synthetic and/or recombinant biomolecules.
  • each read may also comprise information in addition to sequence data (e.g., base-calls), such as estimations of per-position accuracy, features of underlying sequencing technology output (e.g., trace characteristics (integrated counts per peak, shape/height/width of peaks, distance to neighboring peaks, characteristics of neighboring peaks), signal-to-noise ratios, power-to-noise ratio, background metrics, signal strength, reaction kinetics, etc.), and the like.
  • sequence data e.g., base-calls
  • features of underlying sequencing technology output e.g., trace characteristics (integrated counts per peak, shape/height/width of peaks, distance to neighboring peaks, characteristics of neighboring peaks), signal-to-noise ratios, power-to-noise ratio, background metrics, signal strength, reaction kinetics, etc.
  • the sequence reads 1 16 may be generated using essentially any technology capable of generating sequence data from biomolecules, e.g., Maxam- Gilbert sequencing, chain-termination methods, PCR-based methods, hybridization- based methods, ligase-based methods, microscopy-based techniques, sequencing-by- synthesis (e.g., pyrosequencing, SMRT® sequencing, SOLiDTM sequencing (Life Technologies), semiconductor sequencing (Ion Torrent Systems), tSMSTM sequencing (Helicos Biosciences), lllumina® sequencing (lllumina, Inc.), nanopore-based methods (e.g., BASETM, MinlONTM, STRANDTM), etc.).
  • Maxam- Gilbert sequencing e.g., Maxam- Gilbert sequencing, chain-termination methods, PCR-based methods, hybridization- based methods, ligase-based methods, microscopy-based techniques, sequencing-by- synthesis (e.g., pyrosequencing, SMRT® sequencing, SOLiDTM sequencing
  • the sequence information analyzed may comprise replicate sequence information.
  • replicate sequence reads may be generated by repeatedly sequencing the same molecules, sequencing templates comprising multiple copies of a target sequence, sequencing multiple individual biomolecules all of which contain the sequence of interest or "target" sequence, or a combination of such approaches.
  • Replicate sequence reads need not begin and end at the same position in a biomolecule sequence, as long as they contain at least a portion of the target sequence.
  • a circular template can be used to generate replicate sequence reads of a target sequence by allowing a polymerase to synthesize a linear concatemer by continuously generating a nascent strand from multiple passes around the template molecule.
  • Replicate sequences generated from a single template molecule are particularly useful for determining a consensus sequence for that template molecule.
  • This "single-molecule consensus" determination is distinct from the conventional methods for determining consensus sequences from reads of multiple template molecules, and is particularly useful for identifying rare variants that might otherwise be missed in a large pool of sequence reads from multiple templates. Examples of methods of generating replicate sequence information from a single molecule are provided, e.g., in U.S. Patent No. 7,476,503; U.S. Patent Publication No. 20090298075; U.S. Patent Publication No. 20100075309; U.S. Patent Publication No. 20100075327; U.S. Patent Publication No. 20100081 143, U.S.S.N.
  • the accuracy of the sequence read data initially generated by a sequencing technology discussed above may be approximately 70%, 75%, 80%, 85%, 90%, or 95%. Since efficient string graph construction preferably uses high- accuracy sequence reads, e.g., preferably at least 98% accurate, where the sequence read data generated by a sequencing technology has a lower accuracy, the sequence read data may be subjected to further analysis, e.g., overlap detection, error correction etc., to provide the sequence reads 1 16 for use in the string graph generator 1 12. For example, the sequence read data can be subjected to a pre-assembly step to generate high-accuracy pre-assembled reads, as further described elsewhere herein.
  • sequence read data is used to create "pre-assembled reads" having sufficient quality/accuracy for use as sequence reads 1 1 6 in the string graph generator 1 12 to construct the string graph 1 18.
  • a pre-assembly sequence aligner (which may also be referred to as an aggregator) may perform pre-assembly of the sequence read data generated from a sequencing technology (e.g., SMRT® Sequencing or nanopore-based sequencing) to provide the sequence reads 1 16.
  • a sequencing technology e.g., SMRT® Sequencing or nanopore-based sequencing
  • the pre- assembly sequence aligner is very efficient, and certain preferred aligners/aggregators and embodiments for generating pre-assembled reads are described in detail in U.S. Patent Application Nos. 13/941 ,442, filed July 12, 2013; 61/784,219, filed March 14, 2013; and 61/671 ,554, filed July 13, 201 2, which are incorporated herein by reference in their entireties for all purposes.
  • the alignment and consensus algorithm used during pre-assembly is preferably fast, e.g., using simple sorting and counting.
  • the alignment operation comprises choosing a best-match sequence read from the nucleic acid sequence read data as a seed sequence, followed by aligning remaining reads in the sequence read data to the seed sequence to generate the set of pre-assembly aligned sequences.
  • a set of sequence reads for a region of interest or
  • target region (optionally from a mixed population) is generated or otherwise provided, and these sequence reads (e.g., preferably in a FASTA file) are aligned to one another to form a set of sequence alignments.
  • a set of “seed” sequence reads is selected and these seed reads are typically selected from the longest sequence reads in the set, e.g., reads that are at least 3, 4, 5, 6, 8, or 10 kb in length. All the sequence reads in the set are aligned against each of the seed reads, to generate a set of alignments between the reads and the seed reads and, thereby, map each of the reads in the set to at least one seed read.
  • An alignment-and-consensus process is used to construct single "pre-assembled long reads" for each of the seed reads using all of the reads that map to that seed read.
  • the set of sequence alignments generated with the seed read is normalized and used to construct a sequence alignment graph (SAG) analogous to multiple sequence alignment.
  • SAG sequence alignment graph
  • a consensus sequence for the set of sequence reads mapping to that seed read is derived from the SAG, and this consensus sequence can be thought of as representing the "average" sequence of the reads from the mixed population that map to that seed read.
  • those seed reads and all the sequences that map thereto can be combined in a single alignment to derive a single consensus sequence for a resulting pre-assembled long read.
  • pre-assembly is executed using an algorithm based on encoding multiple sequence alignments with a directed acyclic graph to find the best path for the best consensus sequence, and this method is an effective strategy for removing random insertion and missing errors that were present in the original sequence reads.
  • the sequence reads in the sequence alignment graph are partitioned or "clustered” based upon the structure of the graph to generate a plurality of subsets of the set of sequence reads.
  • the constituent sequence reads are aligned and used to construct a sequence alignment graph, which is used to generate a consensus sequence.
  • the new consensus sequences are compared (e.g., by alignment and standard statistical analysis) to reference sequences to identify the source of the sequence reads of the subset of sequence reads from which the consensus sequence was derived.
  • a consensus sequence for a subset may be compared to multiple different reference haplotype sequences for a genomic region of interest, and the reference sequence that best matches the subset consensus sequence is indicative of the haplotype of the original template nucleic acid that was sequenced to generate the sequence reads in the subset.
  • This embodiment is particularly useful for resolving SNP-level diploid sequence variants during the pre-assembly step.
  • the accuracy of the consensus sequence is typically at least 99%, and often at least 99.5%.
  • these highly-accurate consensus sequences are suitable to serve as an input (e.g., sequence reads 1 16) to the string graph assembly method described here.
  • sequence reads 1 16 are provided, they are subjected to alignment and overlap detection by the sequence aligner/overlapper 1 1 0, which generates aligned sequences 1 17.
  • sequence aligner/overlapper 1 10 is very efficient and fast, e.g., using simple sorting and counting, and certain preferred aligners/aggregators are known in the art and/or described with respect to the pre-assembly step, above.
  • the string graph generator 1 12 generates the string graph 1 18 from the aligned sequences
  • Figures 3A and 3B are diagrams illustrating embodiments of methods for creating a string graph from overlaps between aligned sequences and an algorithm for transitive reduction.
  • the string graph generator 1 12 may generate the string graph
  • edges 300 by constructing edges 300 from the aligned, overlapping sequences 1 17 based on where the reads overlap one another.
  • the core of the string graph algorithm is to convert each "proper overlap" between two aligned sequences into a string graph structure.
  • two overlapping reads (aligned sequences 1 17) are provided to illustrate the concepts of vertices and edges with respect to overlapping reads. Specifically, the vertices right at the boundaries of an overlap are g:E and f:E are identified as the "in- vertices" of the new edges to be constructed.
  • Edges 301 are generated by extending from the in-vertices to the ends of the non-overlapping parts of the aligned reads, which are identified as the "out-vertices," e.g., f:E to g:B (out-vertex) and g:E to f:B (out-vertex). If the sequence direction is the same as the direction of the edges, the edge is labeled with the sequence as it is in the sequence read. If the sequence direction is opposite that of the direction of the edges, the edge is labeled with the reverse complement of the sequences.
  • the four aligned, overlapping reads 302 are used to create an initial graph 304, and the initial graph 304 is subjected to transitive reduction 306 and graph reduction, e.g., by "best overlapping," to generate the string graph 1 18.
  • Detecting overlaps in the aligned sequences 1 17 may be performed using overlap-detection code that functions quickly, e.g., using k-mer-based matching.
  • Converting the overlapping reads 302 into the initial graph 304 may comprise identifying vertices that are at the edges of an overlapping region and extending them to the ends of the non-overlapped parts of the overlapping fragments. Each of the edges (shown as the arrows in initial graph 304) is labeled depending on the direction of the sequence. Thereafter, redundant edges are removed by transitive reduction 306 to yield the string graph 1 1 8. Further details on string graph construction are provided in Myers, E. W. (2005) Bioinformatics 21 , suppl. 2, pgs. N79-ii85, which is incorporated herein by reference in its entirety for all purposes.
  • Figure 4 is a diagram graph illustrating an exemplary string graph 400 generated for a double-stranded haploid sample, e.g., E. coli genome, using 10X 1 0,000 base pair (bp) reads, resulting in a string graph comprising 9278 nodes and 9536 edges.
  • a double-stranded haploid sample e.g., E. coli genome
  • 10X 1 0,000 base pair (bp) reads resulting in a string graph comprising 9278 nodes and 9536 edges.
  • Figure 5 is a diagram showing that additional features observed in string graph structures 500 may include areas of entanglement such as branching, knots, and bubbles.
  • Branching or branch points are typically caused by the presence of repeated sequences 502 in the aligned sequences 1 17, but can also be due to the presence of homologous sequences, e.g., where the sample is diploid, and chimeras in the sequence read data can also mimic a repeat region creating an unnecessary branch in the graph.
  • Knots can be caused when an overlap region falls fully within a repetitive region.
  • a simple "best overlapping rule" is typically used to "de-tangle" the knots.
  • Figure 6 is a diagram showing results of applying the best overlapping rule on the E. coli string graph 400. As shown, after the best overlapping rule is applied to the string graph with 400, this "de-tangling" will generate two complementary contigs, one forward strand 600 and one reverse strand 602.
  • the diploid contig generator 1 14 identifies unitigs in the string graph and generates a unitig graph (block 202).
  • non-branching unitigs within the string graph are identified to form the unitig graph, where unitigs represent the contigs that can be constructed unambiguously from the string graph and that correspond to the linear paths in the string graph without any branch induced by repeats or sequencing errors.
  • Figure 7 is a diagram graphically illustrating identifying unitigs from the non- branching parts of the string graph 700 to generate a unitig graph 702, which simplifies the initial string graph into the unitig graph with simple paths in which all the edges and a path without any branching nodes are formed into a single edge.
  • Graph traversal is performed on the unitig graph 702 to generate the contigs 704, which are a contiguous set of overlapping sequences, as shown.
  • Flexible graph construction and graph traversal methods are preferred, e.g., and may be implemented in Python or other computer language, as listed elsewhere herein.
  • Figure 8A is a diagram graphically illustrating a string graph 800 having a quasi linear structure and bubbles 802.
  • Simple bubbles 802 may be generally observed in a string graph 800 where there are local structural variations (SV) between haplotypes.
  • simple bubbles 802 can also be caused by errors in the original sequence reads and/or in the consensus determination performed during the pre- assembly of the reads. If the overlap identification step fails to detect a proper overlap 804 between the reads (e.g., due to a structural variation or sequencing error), a bubble 806 will be rendered in the string graph.
  • Figure 9 is a diagram illustrating one challenge of diploid assembly is to distinguish between similar topologies in a string graph caused by two different types of underlying nucleotide sequence structures in a genome.
  • Sequences 900 having different types of nucleotide sequence structures may have string graph representations 902 that have the same local topology and are therefore difficult to distinguish by conventional assemblers, which focus on local topology rather than regional topology extending over a larger portion of the graph.
  • the string graph representations 902 are illustrating that the region indicated by the dark arrow is present multiple times in the sequences used to generate the graph, and that the sequences on either side of it are different (e.g., due to sequence variants, mutations, different locations on a chromosome, etc.).
  • This string graph representation does not distinguish between whether the underlying nucleotide sequence comprises identical sequences at different positions on a single nucleic acid strand (e.g., on a single chromosome strand or fragment thereof), as shown for repeats sequences 904 (also referred to as repeats, R), or comprises identical sequences on different nucleic acid strands, e.g., homologous chromosomes, as shown for identical homologous sequences 906.
  • haplotype 1 and haplotype 2 may be from different homologous chromosomes, e.g., one maternal chromosome and one paternal chromosome, and the dark arrow is indicative of a region of the chromosomes that is identical between the two homologs.
  • the string graph assembly combines the matching regions (e.g., repeats (R) or identical homologous regions (H)) into a single segment in the graph. Therefore, the resulting string graph representation 902 has the same topology regardless of the underlying sequence structure. The determination of the true, underlying sequence structure can be even more difficult to resolve where there is repeating sequence within homologous regions (not shown).
  • regions e.g., repeats (R) or identical homologous regions (H)
  • the string graph representations 902 of both repeat sequences 904 and identical homologous sequences 906 basically have the same the local structure, as shown, which may be one underlying cause of complex bubbles in the string graph. During assembly, it is desirable to distinguish between these two types of nucleotide sequence structures in order to construct a sequence assembly that accurately represents the sequences of the original sample nucleic acid from which the sequence read data was generated.
  • Figure 10 is a diagram graphically illustrating exemplary large and small scale topological features of the unitig graph 1000, which was generated from genomic sequence data from Arabidopsis thaliana, as described in the EXAMPLE.
  • An enlarged portion 1002 of the string graph 1000 shows both bubbles 1006 caused by structural variations between the homologous copies, as well as a branching point 1004 caused by the presence of repeats in the sequence reads.
  • conventional string graph assembly would treat both of these topological features identically, e.g., by breaking the assembly at both positions, because the conventional methods do not analyze the large scale string graph structure, but instead focus only on the small scale, local structure. Without being bound by theory, it is believed that this limitation is a key weakness of conventional string graph assembly, as further explained below.
  • a conventional graph traversal algorithm typically stops extending contigs around the nodes of such complex bubbles in the unitig graph and only identifies non-branching simple paths. Consequently, typical assembly algorithms break the graph at both the branching point 1004 and the complex bubble 1006 where divergent paths are encountered, which often results in a fragmented assembly.
  • One option is to use a greedy graph traversal algorithm, which may traverse the bubbles to generate larger contigs, but these are less likely to be truly representative of the original sample nucleic acid. Due to the limitations of using short reads and limitations of conventional graph traversal algorithm, conventional methods may sometimes require additional reads for the sequence region to increase fold-coverage in order to help resolve the more complex bubbles.
  • the diploid contig generator 1 14 of the exemplary embodiments is capable of distinguishing between different large-scale topologies in a string graph, e.g., complex bubbles caused by repeats or homologous regions, or true branch points, without requiring the use of additional reads.
  • the diploid contig generator 1 14 uses long reads to generate the string graph (FIG. 2, block 200).
  • a single path through the unitig graph is used to define string bundles as well as corresponding primary contigs. This is accomplished by traversing the unitig graph to identify a set of edges that form non-branching compound paths, referred to herein as "string bundles," e.g., that may contain sequences from multiple haplotypes. Paths that branch from the primary contigs and then rejoin the primary contigs may be designated as associated contigs and are used to define bubble regions of the string bundles.
  • bubble regions are first identified as compound paths in the string graph, which means that this implementation is not constrained by first attempting to find one path through the graph.
  • a new unitig graph is then generated in which each of the compound paths is replaced by a compound edge and each set of simple paths connecting a pair of compound paths in the original unitig graph are replaced in the new unitig graph with a simple edge. This new unitig graph is used to find the primary and associated contigs.
  • Figure 1 1 is a diagram graph illustrating the processing of string bundles, which comprise bubbles as well as "non-bubble" portions.
  • the process may include analyzing each of the string bundles 1 100 to determine a primary contig 1 102 for each string bundle 1 100 (FIG. 2, block 206).
  • determining the primary contig comprises assigning edges in the corresponding string bundle to the primary contig that form a contiguous, end-to-end "best path" sequence that extends the length of the string bundle.
  • a primary contig is a path through a string bundle that explains most of the read overlaps and may represent the sequence of a particular strand of the sample nucleic acid used to generate the sequence read data.
  • Rules for traversing the graph to find the best paths for the contigs can be determined by the ordinary practitioner based on well-established statistical models and methods.
  • Associated contigs 1 1 04 that comprise structural variations as compared to the primary contigs 1 102 are also determined (FIG. 2, block 208).
  • determining the associated contigs comprises assigning edges in paths parallel to the primary contig 1 102 in bubble regions of the string bundle 1 100 as the associated contigs.
  • associated contigs 1 104 may represent sequences that differ between two homologous sequences. The associated contigs 1 104 may be constructed iteratively along the path of the corresponding primary contig 1 102, and the process continues until every edge in the string bundle 1 1 00 is associated with either one of the primary contigs 1 102 or one of the associated contigs 1 1 04. The result of this process is that the string bundle 1 1 00 comprises the primary contigs 1 1 02 plus the locally associated contigs 1 1 04.
  • the contigs in each of the string bundles 1 100 are analyzed to distinguish junctions in the respective string bundles caused by the presence of homologous regions having structural variations from those that indicate true branching paths, e.g., caused by the presence of repeat sequences 904 within a nucleic acid sequence.
  • the contigs are analyzed to identify candidate branch points in the primary contigs 1 102.
  • the primary contigs are broken at these branch points to provide corrected primary contigs 1 1 02 along with their locally associated contigs 1 104.
  • One aspect of the exemplary embodiments is the recognition of the importance of distinguishing a junction in a unitig graph as a vertex belonging to a string bundle or a vertex of a branching path from which a primary contig 1 102 and associated contigs 1 1 04 diverge. Consequently, the diploid contig generator 1 14 determines whether the vertex is indicative of minor structural variation between two homologous sequences that can remain within the string bundle, or indicative of a major structural topology resulting in a branching path that cannot remain within the string bundle and requires the assembly be broken at that point.
  • Figure 12 is a diagram illustrating a process for determining whether a junction at vertex 1202 in a unitig graph 1200 belongs to a string bundle 1204 or is indicative of a branching path 1206. In one embodiment, this may be accomplished by analyzing a distance at which two downstream paths of a vertex rejoin, where one of the paths may define a primary contig 1208 and the other path may define a candidate associated contig 1210. For example, given a junction at vertex U, and two downstream paths V and W, it is determined whether V and W meet within a predefined radius R from the vertex U. If the two downstream paths (e.g., V and W) rejoin within a predefined radius, then the two paths are identified as part of a single string bundle 1204.
  • V and W e.g., V and W
  • the radius is a selectable parameter that may be tunable by the operator, as it depends on the genome structure. As a point of reference, however, the radius may be approximately 10 base calls in length in the EXAMPLE above. In one embodiment, the radius may be selected prior to assembly based on known characteristics (e.g., size) of structural variations in the sample nucleic acids.
  • the length of the radius should be selected to so that the bubbles fully contain the structural variations and allow the two downstream paths of the bubble to rejoin within the radius to avoid breaking the bundle.
  • the results can be used to determine a radius for a subsequently performed assembly.
  • the radius can be increased and the assembly process re-run to try to increase the contig lengths in the final assembly.
  • a radius of a shorter length may be selected.
  • the string bundle is broken at the branch points after the primary contigs and the associated contigs are determined
  • the string bundle may be broken at the branch points at an earlier stage during processing.
  • Embodiment 2 Identifying String Bundles and Determining Primary and Associated Contigs
  • Figure 13 is a diagram graph illustrating the processing of string bundles according to a second embodiment.
  • the goal is to first identify bubble regions as compound paths.
  • One purpose of this is to attempt to decompose the string graph into simple paths and simple bubbles.
  • the string graph for a diploid genome with complicated heterozyguous structure variations or repeat structures may not be easy to decompose into simple path and simple bubbles due to possible subgraph motifs.
  • Step 1 the initial string graph is simplified to a graph, UGo, for example, having simple paths in which edges in a path without any branching node are represented with single edge.
  • step 2 nodes 1303 having multiple out-edges in UGo are found and for each of these nodes, a search is initiated to find a local "bundle" of edges.
  • tracers or labels, are assigned to the nodes 1303 having multiple out-edges to trace down each branch from a source node to a sink node.
  • An assigned tracer may be active or inactive. Finding the local bundles of edges includes the following sub-steps.
  • each node having an active tracer is
  • Loops are detected in response to determining that any offspring node of a parent node that has an active tracer already has an assigned tracer. When a loop is detected, the search stops and no compound path is generated.
  • the number of active tracer can increase quickly. Therefore, only a predefined number of active tracers are assigned. The searches stopped when the number of active tracers assigned exceeds the predefined number.
  • the search is stopped when the number of nodes and the length of exceed predefined threshold.
  • step 3 for compound paths 1 302 that are overlapped with others, or for nested compound paths (e.g., a smaller compound path is part of a larger compound path), the longest compound path is selected and the smaller compound path ignored.
  • a new unitig graph (UGi) 1306 is generated in which each of the compound paths 1 302 identified in UGo, are replaced by a single compound edge 1308; and each the simple paths 1 304 in UGo connecting the compound paths 1302 are replaced with a simple edge 1310.
  • the resulting unitig graph UGi contains compound edges 1308 connected by simple edges 1310 and is used to identify the string bundles, primary contig and associated contigs, as described above.
  • the result of the above processing is a string bundle 1204 comprising corrected primary contigs 122 along with their locally associated contigs 124 (FIG. 1 ).
  • the output from the diploid contig generator 1 14 may include the primary contigs 122, the associated contigs 1 24, and optionally a final assembly graph 126 that comprises the primary and associated contigs.
  • Figure 14 is a diagram graphically illustrating construction of a final consensus sequence based on the primary contigs and the associated contigs.
  • a primary and associated contig assembly 1400 is shown including large structural variations (SV) 1402 (shown as rectangles) and also single-nucleotide polymorphisms (SNPs) and small SVs 1404.
  • Two of the large SVs 1402 belong to one nucleic acid molecule 1408 and the other two large SVs 1402 belong to another nucleic acid molecule 141 0.
  • This provides the general structure that needs to be resolved to provide the sequences for the individual sample nucleic acid molecules 1408 and 1410, e.g., two homologous chromosomes.
  • the diploid contig assembler 1 14 (or other component) may apply logic to determine which alleles go together in a single nucleic acid to provide the haplotype for that molecule.
  • this may be accomplished by examining the allelic constitution the long sequence reads 1 16 (FIG. 1 ) to determine whether a single sequence read contains more than one of these variant positions (large SVs or SNPs). When it is determined that a single read (which is necessarily from a single molecule) comprises loci for more than one of the variant positions, the alleles at those loci are identified as linked, originating from a single original nucleic acid molecule.
  • allelic constitution of the long sequence reads 1 16 has been determined with respect to the primary and associated contig assembly 1400, it can be determined which version of each variant position originates with which nucleic acid molecule 1408 and 1410, and thereby determine the final consensus sequence for the original nucleic acid molecules.
  • the method of the exemplary embodiments is capable of generating long contigs for polyploid genomes that keep the genomic information for a single nucleic acid strand together in a comprehensive and concise data structure/representation, and also allow for haplotype sequencing of homologous chromosomes.
  • Results of the methods and systems disclosed herein may be used for consensus sequence determination from biomolecule sequence data.
  • the system includes a computer-readable medium operatively coupled to the processor that stores instructions for execution by the processor.
  • the instructions may include one or more of the following: instructions for receiving input of sequence reads (and, optionally, reference sequence information), instructions for constructing pre-assembled reads, instructions for aligning sequence reads, instructions for generating string graphs, instructions for generating unitig graphs, instructions for identifying string bundles, instructions for determining primary contigs, instructions for determining associated contigs, instructions for correcting reads, instructions for generating consensus sequences, instructions for generating haplotype sequences, instructions that compute/store information related to various steps of the method (e.g., edges and nodes in a string graph, overlaps and branch points in a string graph, primary and associated contigs, and instructions that record the results of the method.
  • steps of the method e.g., edges and nodes in a string graph, overlaps and branch points in a string graph, primary and associated contigs, and instructions that record the results of the method.
  • the methods are computer-implemented methods.
  • the algorithm and/or results (e.g., consensus sequences generated) are stored on computer-readable medium, and/or displayed on a screen or on a paper print- out.
  • the results are further analyzed, e.g., to identify genetic variants, to identify one or more origins of the sequence information, to identify genomic regions conserved between individuals or species, to determine relatedness between two individuals, to provide an individual with a diagnosis or prognosis, or to provide a health care professional with information useful for determining an appropriate therapeutic strategy for a patient.
  • the computer-readable media may comprise any combination of a hard drive, auxiliary memory, external memory, server, database, portable memory device (CD-R, DVD, ZIP disk, flash memory cards, etc.), and the like.
  • the invention includes an article of manufacture for string graph assembly of polyploid genomes that includes a machine-readable medium containing one or more programs which when executed implement the steps of the invention as described herein.
  • the methods described herein were used to perform sequence analysis of the 120 Mb Arabidopsis genome.
  • the strategy comprised generating a "synthetic" diploid dataset by using two inbred strains of Arabidopsis, Ler-0 and Col-0. The two strains were sequenced separately, then sequencing reads generated for each were pooled and subjected to pre-assembly followed by the string graph diploid assembly strategy described herein to determine if this strategy could correctly assemble the two strains from the pooled read data.
  • the sequence reads used as input in the diploid assembly process ranged from about 1 0 kb to about 22 kb, with the majority of the reads between 10 and 15 kb.
  • the unitig graph shown in Figure 10 was constructed from a string graph generated using the pooled sequencing reads. The next step was to find an end- to-end path though the unitig graph along which a string bundle could be built.
  • the compound paths of the string bundle contained sequences from both "haplotypes" (in this case, both strains).
  • the string bundle comprised a primary contig and the locally associated contigs, where the primary contig is the path from the beginning to the end of the string bundle that explains most of the overlaps, and the associated contigs are the paths in parallel to the primary contig in the bubbles present in the string bundle.
  • the process was continued until there were no edges left, and the string bundle was subsequently broken at branching points believed to be caused by repeats to provide corrected primary contigs and locally associated contigs.

Landscapes

  • Life Sciences & Earth Sciences (AREA)
  • Physics & Mathematics (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Health & Medical Sciences (AREA)
  • Engineering & Computer Science (AREA)
  • Biotechnology (AREA)
  • Medical Informatics (AREA)
  • Biophysics (AREA)
  • Theoretical Computer Science (AREA)
  • Spectroscopy & Molecular Physics (AREA)
  • Proteomics, Peptides & Aminoacids (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Chemical & Material Sciences (AREA)
  • Evolutionary Biology (AREA)
  • General Health & Medical Sciences (AREA)
  • Analytical Chemistry (AREA)
  • Molecular Biology (AREA)
  • Genetics & Genomics (AREA)
  • Measuring Or Testing Involving Enzymes Or Micro-Organisms (AREA)

Abstract

Selon des modes de réalisation donnés en exemples, l'invention concerne des procédés et des systèmes pour l'assemblage de graphes de chaînes de génomes polyploïdes. Des aspects du mode de réalisation donné en exemple comprennent la réception d'un graphe de chaînes généré à partir de lectures de séquences d'une longueur d'au moins 0,5 kb ; l'identification d'unitigs dans le graphe de chaînes et la génération d'un graphe d'unitigs ; et l'identification de groupes de chaînes dans le graphe d'unitigs par : détermination d'un contig primaire dans chacun des groupes de chaînes ; et la détermination de contigs associés qui contiennent des variations structurales comparativement au contig primaire.
PCT/US2016/038264 2015-06-18 2016-06-18 Assemblage de graphes de chaînes pour génomes polyploïdes WO2016205767A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
US14/743,713 US20150286775A1 (en) 2013-12-18 2015-06-18 String graph assembly for polyploid genomes
US14/743,713 2015-06-18

Publications (1)

Publication Number Publication Date
WO2016205767A1 true WO2016205767A1 (fr) 2016-12-22

Family

ID=57546607

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/US2016/038264 WO2016205767A1 (fr) 2015-06-18 2016-06-18 Assemblage de graphes de chaînes pour génomes polyploïdes

Country Status (1)

Country Link
WO (1) WO2016205767A1 (fr)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108753765A (zh) * 2018-06-08 2018-11-06 中国科学院遗传与发育生物学研究所 一种构建超长连续dna序列的基因组组装方法
CN112786110A (zh) * 2021-01-29 2021-05-11 武汉希望组生物科技有限公司 一种序列组装方法及系统
CN113496760A (zh) * 2020-04-01 2021-10-12 深圳华大基因科技服务有限公司 基于第三代测序的多倍体基因组组装方法和装置

Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140025312A1 (en) * 2012-07-13 2014-01-23 Pacific Biosciences Of California, Inc. Hierarchical genome assembly method using single long insert library

Patent Citations (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140025312A1 (en) * 2012-07-13 2014-01-23 Pacific Biosciences Of California, Inc. Hierarchical genome assembly method using single long insert library

Non-Patent Citations (5)

* Cited by examiner, † Cited by third party
Title
CAO ET AL.: "De novo assembly of a haplotype-resolved human genome.", NATURE BIOTECHNOLOGY, vol. 33, no. 6, 25 May 2015 (2015-05-25), pages 617 - 622, XP055334053 *
HENSON ET AL.: "Next-generation sequencing and large genome assemblies.", PHARMACOGENOMICS., vol. 13, no. 8, June 2012 (2012-06-01), pages 901 - 915, XP055334054 *
HUANG ET AL.: "HaploMerger: Reconstructing allelic relationships for polymorphic diploid genome assemblies .", GENOME RESEARCH, vol. 22, no. 8, August 2012 (2012-08-01), pages 1581 - 1588, XP055334052 *
IQBAL ET AL.: "De novo assembly and genotyping of variants using colored de Bruijn graphs.", NATURE GENETICS, vol. 44, no. 2, February 2012 (2012-02-01), pages 226 - 232, XP055334060 *
SIMPSON ET AL.: "Efficient construction of an assembly string graph using the FM-index.", BIOINFORMATICS, vol. 26, no. 12, 15 June 2016 (2016-06-15), pages I307 - I373, XP055334061 *

Cited By (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108753765A (zh) * 2018-06-08 2018-11-06 中国科学院遗传与发育生物学研究所 一种构建超长连续dna序列的基因组组装方法
CN108753765B (zh) * 2018-06-08 2020-12-08 中国科学院遗传与发育生物学研究所 一种构建超长连续dna序列的基因组组装方法
CN113496760A (zh) * 2020-04-01 2021-10-12 深圳华大基因科技服务有限公司 基于第三代测序的多倍体基因组组装方法和装置
CN113496760B (zh) * 2020-04-01 2024-01-12 深圳华大基因科技服务有限公司 基于第三代测序的多倍体基因组组装方法和装置
CN112786110A (zh) * 2021-01-29 2021-05-11 武汉希望组生物科技有限公司 一种序列组装方法及系统
CN112786110B (zh) * 2021-01-29 2023-08-15 武汉希望组生物科技有限公司 一种序列组装方法及系统

Similar Documents

Publication Publication Date Title
US10783984B2 (en) De novo diploid genome assembly and haplotype sequence reconstruction
US20150286775A1 (en) String graph assembly for polyploid genomes
US11817180B2 (en) Systems and methods for analyzing nucleic acid sequences
US20150169823A1 (en) String graph assembly for polyploid genomes
US20210217490A1 (en) Method, computer-accessible medium and system for base-calling and alignment
US10839940B2 (en) Method, computer-accessible medium and systems for score-driven whole-genome shotgun sequence assemble
US20130317755A1 (en) Methods, computer-accessible medium, and systems for score-driven whole-genome shotgun sequence assembly
US10777301B2 (en) Hierarchical genome assembly method using single long insert library
US10229519B2 (en) Methods for the graphical representation of genomic sequence data
US20130138358A1 (en) Algorithms for sequence determination
US20110257889A1 (en) Sequence assembly and consensus sequence determination
JP2022533801A (ja) 合成による高速フォワードシークエンシング
Blanchette Computation and analysis of genomic multi-sequence alignments
US20180060484A1 (en) Extending assembly contigs by analyzing local assembly sub-graph topology and connections
CN113963749A (zh) 高通量测序数据自动化组装方法、系统、设备及存储介质
WO2016205767A1 (fr) Assemblage de graphes de chaînes pour génomes polyploïdes
Rachappanavar et al. Analytical Pipelines for the GBS Analysis
Ebler Design and application of methods for genome inference
WO2018033733A1 (fr) Procédés et appareil permettant d'identifier des variants génétiques
US20220284986A1 (en) Systems and methods for identifying exon junctions from single reads
Bolognini Unraveling tandem repeat variation in personal genomes with long reads
Ning et al. Next‐Generation Sequencing Technologies and the Assembly of Short Reads into Reference Genome Sequences
Masoudi-Nejad et al. De novo assembly algorithms
Girilishena Complete computational sequence characterization of mobile element variations in the human genome using meta-personal genome data
Tarafder Probabilistic method for filling gaps in genome assemblies

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 16812596

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 16812596

Country of ref document: EP

Kind code of ref document: A1