Machine learning models for recalibrating genotype calls from existing sequencing data files

Abstract

This disclosure describes a method, a non-temporal computer-readable medium, and a system that can recalibrate genotype calls (e.g., mutation calls) in existing sequencing data files using a machine learning model. For example, the disclosed system may have access to one or more existing sequencing data files for a genomic sample, the files containing nucleotide read data and genotype calls at specific genomic coordinates. From one or more existing sequencing data files, the disclosed system extracts sequencing metrics for nucleotide reads or specific genotype calls at specific genomic coordinates. By processing the extracted sequencing metrics, the system further generates mutation call classifications indicating the accuracy of specific genotype calls using a call recalibration machine learning model. In some cases, the system updates or recalibrates genotype calls or quality measurement sequencing metrics for genotype calls based on the mutation call classifications.

Description

Technical Field

[0001] (Cross - reference to related applications) This application claims the priority and benefit of U.S. Provisional Patent Application No. 63 / 499,845, filed on May 3, 2023, entitled "MACHINE LEARNING MODEL FOR RECALIBRATING GENOTYPE CALLS FROM EXISTING SEQUENCING DATA FILES", the entire content of which is incorporated herein by reference.

Background Art

[0002] In recent years, biotechnology companies and research institutions have been improving the hardware and software for sequencing nucleotides and determining variant calls for genomic samples. For example, some existing nucleotide - based sequencing platforms determine individual nucleotide bases within a sequence by using traditional Sanger sequencing or by using the sequencing - by - synthesis (SBS) method. When using SBS, existing platforms can monitor thousands to millions of nucleic acid polymers synthesized in parallel to predict nucleotide base calls from larger base call data sets. For example, cameras in many SBS platforms capture images of irradiated fluorescent tags incorporated into oligonucleotides to determine nucleotide base calls. After capturing such images, existing SBS platforms transmit the base call data (or image - based data) to a computing device and apply sequencing data analysis software to determine the nucleotide base sequence of the nucleic acid polymer. Based on the differences between the aligned nucleotide reads and a reference genome, existing systems can further utilize variant callers to identify variants of genomic samples such as single nucleotide polymorphisms (SNPs), insertions or deletions (indels), and / or structural variations, as well as genotype calls.

[0003] Despite these recent advances in sequencing and mutation calling, existing nucleotide-based sequencing platforms and sequencing data analysis software (collectively, existing sequencing systems) often perform mutation calls that inaccurately determine nucleotide-based calls or corresponding mutation calls, requiring significant computational resources to determine or update models for mutation calls. For example, some existing sequencing systems inefficiently consume considerable computing resources by using overly complex models (often requiring substantial computing runtime) to accurately determine base calls or mutation calls. For instance, some existing sequencing systems utilize mutation calls with deep learning architectures or some other neural network architecture that require large computational resources (e.g., computation time, processing power, and memory) to train and apply. For example, some existing sequencing systems utilize deep learning architectures that require significant time across multiple computing devices to generate genotype calls for a single sample sequence, even after training.

[0004] In contrast to some deep learning architectures, several machine learning-based mutation colas have been developed that accurately determine mutation calls by processing various features on both nucleotide reads and reference genomes. However, some such machine learning-based mutation colas are limited to certain types of processors. For example, some existing machine learning-based mutation colas can only run on field programmable gate arrays (FPGAs) or similar processing systems. Due to such technical limitations, such existing machine learning-based mutation colas often cannot run on remote servers running more mainstream processors, such as one or more of a central processing unit (CPU) or a graphical processing unit (GPU).

[0005] However, as machine learning-based mutation colas are updated or newly introduced, some existing sequencing systems rearrange or regenerate mutation call files for existing sequencing datasets to take advantage of the improved accuracy of the newer mutation colas. While rearranging or regenerating mutation call files using updated or newer machine learning-based mutation colas often improves mutation calling accuracy, such rearrangement or regeneration approaches can double or triple the computational resources required to determine accurate mutation calls for a given genomic sample. For example, for a database of over 100,000 genomic samples running on a previous version of the mutation calling model, the computational runtime for a newer version of machine learning-based mutation colas to determine mutation calls from read data could reach 2,000,000 hours at 20 minutes per genomic sample.

[0006] Despite the computational time and other resources required to run machine learning-based mutation colas on previously called genomic samples, some existing sequencing systems are worth investing such resources in because the mutation calling accuracy of such systems yields significant diagnostic or research outcomes. In some situations, for example, existing sequencing systems apply mutation colas that inaccurately identify an excessive number of false-negative mutation calls. For example, an existing sequencing system might determine that a genomic coordinate represents a homozygous reference genotype (and therefore does not contain the mutation) when the coordinate actually does contain a mutation. Indeed, while existing mutation colas achieve a certain level of accuracy, their limitations mean there is still room for improvement in recovering false-negative mutation calls. To illustrate the impact of such inaccuracies, a mutation call identifying a specific single nucleotide polymorphism (SNP) in the hemoglobin beta (HBB) gene can have significant implications. For example, if a mutation colas identifies a SNP at rs344 on chromosome 11, the mutation colas may either accurately identify the genetic cause of sickle cell anemia or miss the cause of the disease. As a further example, mutation calls that accurately or inaccurately identify deletions of one or more copies of the hemoglobin subunit alpha1 (HbA1) or hemoglobin subunit alpha2 (HbA2) gene can either accurately identify the genetic cause of an inherited blood disorder or completely miss the gene deletion.

[0007] One contributing factor to the aforementioned inaccuracies is that many existing sequencing systems utilize only a limited set of data when determining nucleotide-based calls. For example, existing sequencing systems often rely exclusively on information directly extracted from nucleotide reads of a sample sequence, such as read depth, mismatch count, sequence alignment score, and mapping quality, to determine nucleotide-based calls. While sequencing information from nucleotide reads can provide valuable insights into determining nucleotide-based calls, existing sequencing systems that rely solely on this data may have reduced performance in accurately determining nucleotide-based calls, including mutation calls. In fact, some existing sequencing systems that rely on raw sequence data inaccurately determine SNPs, indels, or other mutations in genomic sample sequences compared to more complex models. Indeed, existing sequencing systems often identify false-negative or false-positive mutations in the U.S. Food and Drug Administration (FDA) Truth Challenges. [Overview of the Initiative]

[0008] This disclosure describes a method, a non-temporal computer-readable medium, and embodiments of a system that may utilize a machine learning model to recalibrate genotype calls (e.g., mutation calls) or mutation quality metrics in an existing sequencing data file. As described below, the disclosed system may access one or more existing sequencing data files for a genomic sample, such existing files containing nucleotide read data and genotype calls at specific genomic coordinates. From such existing sequencing data files, the disclosed system extracts sequencing metrics for nucleotide reads or specific genotype calls at specific genomic coordinates. By processing the extracted sequencing metrics, the system further generates mutation call classifications indicating the accuracy of specific genotype calls, utilizing a call recalibration machine learning model. Based on the mutation call classifications, the system may optionally update or recalibrate specific genotype calls or quality measurement sequencing metrics corresponding to specific genotype calls. After recalibrating specific and other genotype calls for a genomic sample, the disclosed system may output an updated or recalibrated sequencing data file for the genomic sample, e.g., an updated mutation call file.

[0009] By extracting sequencing metrics for genotype calls from existing sequencing data files and utilizing a call recalibration machine learning model to generate updated, previously generated genotype calls or variant call classifications for the corresponding metrics, the disclosed system can improve accuracy, efficiency, and speed compared to existing sequencing systems. Importantly, the disclosed system can also avoid the computational cost of re-running the entire machine learning-based variant call model on sequencing data for genomic samples to determine updated genotype calls or updated quality measurement sequencing metrics for genomic samples, as some existing sequencing systems must do. [Brief explanation of the drawing]

[0010] The "modes for carrying out the invention" refer to the drawings briefly described below. [Figure 1] A block diagram of a sequencing system including a call recalibration system, according to one or more embodiments, is shown. [Figure 2A] This document outlines, in one or more embodiments, (i) an original call generation model that processes base call data for a genomic sample to generate an original sequencing data file, and (ii) an updated call generation model that reprocesses the base call data for the genomic sample through a call recalibration machine learning model to generate an updated sequencing data file. [Figure 2B] This document outlines a call recalibration system that uses one or more call recalibration machine learning models to analyze information from the original sequencing data file and generate a recalibrated sequencing data file. [Figure 3] This document describes a call recalibration system, in one or more embodiments, that receives an existing sequencing data file, extracts sequencing metrics from it, generates a variant call classification based on the extracted sequencing metrics, and generates a recalibrated sequencing data file (e.g., an updated variant call file) based on the variant call classification. [Figure 4A-C] This document describes a call recalibration system, according to one or more embodiments, that identifies or extracts sequencing metrics from existing sequencing data files or external sources and generates a variant call classification. [Figure 5A] This invention provides a call recalibration system that generates variant call classifications (e.g., genotype probabilities) in one or more embodiments, uses a call recalibration machine learning model to generate corresponding recalibrated genotype calls or recalibrated sequencing metrics, and generates a recalibrated genotype call data file containing updated genotype calls based on such classifications. [Figure 5B]This invention provides a call recalibration system that generates variant call classifications (e.g., genotype probabilities) in one or more embodiments, uses a call recalibration machine learning model to generate corresponding recalibrated genotype calls or recalibrated sequencing metrics, and generates a recalibrated genotype call data file containing updated genotype calls based on such classifications. [Figure 5C] This invention provides a call recalibration system that generates variant call classifications (e.g., genotype probabilities) in one or more embodiments, uses a call recalibration machine learning model to generate corresponding recalibrated genotype calls or recalibrated sequencing metrics, and generates a recalibrated genotype call data file containing updated genotype calls based on such classifications. [Figure 6] This document describes an exemplary process for a call recalibration system that trains a call recalibration machine learning model, according to one or more embodiments. [Figure 7] The table shows the computational nodes and runtimes for reprocessing base call data for a genomic sample using an updated call generation machine learning model, compared to recalibrating genotype calls for a genomic sample using a call recalibration machine learning model, according to one or more embodiments. [Figure 8A] The graphs show the results of reprocessing sequencing data using existing call generation models in one or more embodiments, and the results of recalibrating existing genotype calls using call recalibration machine learning models, comparing the results of false positives and false negatives (for SNPs or indels). [Figure 8B] The graphs show the results of reprocessing sequencing data using existing call generation models in one or more embodiments, and the results of recalibrating existing genotype calls using call recalibration machine learning models, comparing the results of false positives and false negatives (for SNPs or indels). [Figure 9] A flowchart shows a series of operations for generating a recalibrated sequencing data file using one or more embodiments. [Figure 10]An exemplary block diagram of a computing device for performing one or more embodiments of this disclosure is shown. [Modes for carrying out the invention]

[0011] This disclosure describes embodiments of a call recalibration system that utilizes a call recalibration machine learning model to recalibrate genotype calls (e.g., variant calls) or corresponding sequencing metrics for a sample nucleotide sequence. To recalibrate such genotype calls or corresponding metrics, the call recalibration system accesses one or more existing sequencing data files for a genomic sample. Such sequencing data files may include, for example, genotype call data files such as variant call format (VCF) files, and / or alignment data files such as binary alignment map (BAM) files or compressed reference-oriented alignment map (CRAM) files. By processing sequencing metrics extracted from existing sequencing data files through a call recalibration machine learning model, the call recalibration system may generate variant call classifications or predictions to verify, recalibrate, or correct genotype calls previously generated by a call generation model. Based on variant call classification or prediction, the call recalibration system may also verify or update various sequencing metrics of previously generated genotype calls, such as call quality, genotype associated with the call, genotype quality associated with the genotype, Phred-scaled Likelihood (PL), and / or other metrics with corresponding fields. By utilizing a call recalibration machine learning model to update sequencing metrics, the call recalibration system may improve the confidence or reliability of updated or verified genotype calls at specific genomic coordinates.

[0012] As described above, in some embodiments, the call recalibration system identifies sequencing metrics for previously generated genotype calls. For example, the call recalibration system extracts or determines sequencing metrics for a sample nucleotide sequence from one or more existing sequencing data files. More specifically, in certain implementations, the call recalibration system extracts or determines different types of sequencing metrics associated with different sources from one or more existing sequencing data files. For example, the call recalibration system extracts or determines read-based sequencing metrics, including metrics derived from nucleotide reads of the sample nucleotide sequence. In some embodiments, for example, the call recalibration system extracts read-based sequencing metrics from sequencing data files containing nucleotide reads of the sample nucleotide sequence, such as alignment data files (e.g., binary alignment map (BAM) files or compressed reference-oriented alignment map (CRAM) files).

[0013] Furthermore, in some embodiments, the call recalibration system extracts or determines call model-generated sequencing metrics generated via mutant cola or other call generation models, such as internal variables of the call recalibration system that are not accessible to other systems or parties (e.g., unique quality score, base context, read filtering, unique hypothesis score, and other metrics). In some implementations, the call recalibration system extracts at least some of the call model-generated sequencing metrics from genotype call data files, such as mutant call format (VCF) files or genomic mutant call format (gVCF) files.

[0014] Beyond read-based or call model-generated sequencing metrics, in some implementations, call recalibration systems derive or reconstruct one or more sequencing metrics, such as by reconstructing one or more call model-generated sequencing metrics that are not explicitly written to (or stored in) the sequencing data file, from other information stored within the sequencing data file. In fact, in some cases, call recalibration systems determine call model-generated sequencing metrics in the form of variant call sequencing metrics and mapping and alignment sequencing metrics, some of which are derived from or determined by one or more existing sequencing data sets in various formats. For example, in some embodiments, instead of relying on hidden Markov model (HMM) scores for each read, call recalibration systems utilize concise idiosyncratic gaped alignment reports (CIGARs) to estimate the alignment of each read in a sequencing data file (e.g., an alignment data file) relative to a reference genome.

[0015] In addition, in some embodiments, the call recalibration system extracts or determines sequencing metrics from external sources identified from one or more external databases showing genomic sequences associated with various nucleotide attributes, mapping challenges, and sequencing biases. Alternatively or additionally, the call recalibration system extracts or determines sequencing metrics from external sources stored in existing sequencing data files.

[0016] By processing the extracted array determination metric or the reconstructed array determination metric, in certain implementations, the call recalibration system generates a set of predicted classifications by which the system can modify or improve a given genotype call or a field associated with a given genotype call. More specifically, in some embodiments, the call recalibration system utilizes a call recalibration machine learning model to generate a set of variant call classifications that affect or reflect the accuracy of identifying variants at specific genomic coordinates from the array determination metric. Depending on the type of genomic coordinates, the call recalibration system utilizes a specifically trained version of the call recalibration machine learning model to generate, for example, (i) variant call classifications for multi-allelic loci, and (ii) variant call classifications different from specific variant call classifications from different versions of the call recalibration machine learning model for haploid coordinates or pseudo-homozygous reference coordinates.

[0017] For example, when generating mutation call classifications for multi-allele genomic coordinates using existing genotype calls, the call recalibration system may use a call recalibration machine learning model to generate a set of mutation call classifications that include (i) a reference probability in which the genotype call contains a homozygous reference genotype in multi-allele genomic coordinates, (ii) a zygosity error probability in which the genotype call contains a zygosity error in multi-allele genomic coordinates, and (iii) a true positive mutation probability in which the genotype call constitutes a true positive mutation in multi-allele genomic coordinates. In another embodiment, for haploid coordinates in existing genotype calls, the call recalibration system may use a call recalibration machine learning model to generate a set of mutation call classifications that include (i) a first genotype probability for a first genotype in genomic coordinates and (ii) a second genotype probability for a second genotype in genomic coordinates. Furthermore, for reference coordinates that are likely to be homozygous, the call recalibration system may use a call recalibration machine learning model to generate a set of variant call classifications that include (i) the false positive probability that the genotype call is a false positive mutation, (ii) the zygosity error probability that the genotype call includes a zygosity error (e.g., the probability that it identifies the correct alternative allele but includes a zygosity error (e.g., 0 / 1 instead of 1 / 1, or 1 / 1 instead of 0 / 1), or the probability that it misidentifies the genotype of the nucleotide-based call), and (iii) the true positive probability (e.g., a homozygous alternative classification that shows the probability that the genotype call contains a true positive mutation).

[0018] Based on the variant call classification, the call recalibration system can identify, correct, or update genotype calls or sequencing metrics corresponding to one or more genotype calls for genomic coordinates (e.g., variant calls or non-variant calls). For example, the call recalibration system can utilize the variant call classification to update genotype data fields within a genotype call data file (e.g., a variant call format file or other base call output file) that indicate or represent updated genotype calls with improved accuracy. As shown above, in certain implementations, the call recalibration system can update genotype calls for specific genomic coordinates such as bi-allelic genomic coordinates, haploid genomic coordinates, and / or homozygous reference coordinates that may be inaccurately identified (e.g., genomic coordinates previously inaccurately identified by a variant caller as indicating a homozygous reference genotype).

[0019] As a further example, in some implementations, the call recalibration system can utilize (i) sequencing metrics extracted or determined from one or more existing sequencing data files and (ii) a call recalibration machine learning model to modify data fields corresponding to an existing variant call file (or other genotype call data file) for genotype calls. For example, the call recalibration system can update one or more of the base call quality metric, genotype probability metric, genotype likelihood metric, or genotype quality metric for a genotype call within a corresponding field of a VCF or other sequencing data file. If the call recalibration system determines that a modified base call quality metric (e.g., Q-score) or other confidence score does not meet a threshold, the call recalibration system can annotate the genotype call within the recalibrated VCF or other recalibrated sequencing data file to indicate that the modified metric or score is below a base call quality threshold or other metric or score threshold.

[0020] As suggested above, the call recalibration system offers several advantages, benefits, and / or improvements over existing sequencing systems, including mutant cola and other sequencing data analysis software. For example, the call recalibration system verifies or corrects genotype calls (or corresponding sequencing metrics) with less computational runtime than existing sequencing systems that implement machine learning-based mutant cola. By extracting sequencing metrics from existing sequencing data files to analyze genotype calls and associated sequencing data, the call recalibration system significantly improves processing runtime compared to existing sequencing systems that reanalyze nucleotide read data using new or updated call-generating machine learning models. In some implementations, for example, the call recalibration system shows a 65% reduction in runtime per genome sample sequencing compared to reanalyzing corresponding sequencing data using existing call-generating machine learning models. The disclosure further illustrates such improved computational runtime below, at least with respect to Figures 2A, 2B, and 7.

[0021] The improved efficiency and speed of call recalibration systems are particularly evident for machine learning-based mutation calling employing deep learning architectures. As mentioned above, some existing sequencing systems utilize computationally expensive and slow neural network architectures (e.g., deep learning architectures such as convolutional neural networks) that require a lot of time (e.g., 5-8 hours on multiple processors running on a server) and significant computational resources, even just to implement and generate files using mutation calls from the sequencing process. Such deep learning architectures may even require days (or weeks) to train. Conversely, call recalibration systems utilize relatively lightweight and fast architectures for call recalibration machine learning models. In contrast to the much time across multiple processors required by existing sequencing systems, call recalibration systems often require less than 10 minutes of runtime on a general-purpose CPU to recalibrate nucleotide-based calls to a sample nucleotide sequence. Thus, call recalibration systems are far faster and less computationally expensive than many deep learning approaches to mutation calling.

[0022] In addition to rapid computing processing, in some embodiments, call recalibration systems increase processing flexibility, allowing sequencing systems to use machine learning models to determine, modify, or update genotype calls or corresponding sequencing metrics. As shown above, some existing machine learning-based mutant calls run exclusively on, for example, field-programmable gate arrays (FPGAs) or other hardware accelerators. In contrast, in one or more embodiments, the call recalibration machine learning model of a call recalibration system may run on a general-purpose processing unit, such as one or more of a central processing unit (CPU) or a graphical processing unit (GPU), but not limited to these. By training a call recalibration machine learning model to modify and / or update genotype calls in existing sequencing data files, a call recalibration system can be implemented with significantly fewer computing resources compared to existing sequencing systems that utilize a call generation machine learning model to generate or regenerate genotype calls. Thus, a call recalibration system can also be implemented with fewer processing cores and less processing memory. In one or more implementations, for example, a call recalibration system requires one-third the processing cores and half the processing memory compared to reanalyzing using existing machine learning-based mutant colas, but exhibits the faster runtime discussed above, resulting in cost reductions. This disclosure further demonstrates such improved processing flexibility below, at least with respect to Figure 7.

[0023] Beyond improved runtime and flexibility, the call recalibration system improves the accuracy of genotype calls or corresponding sequencing metrics from existing sequencing data files for genomic samples. In one or more embodiments, for example, the call recalibration system may use a call recalibration machine learning model to update genotype calls, or corresponding base call quality metrics, genotype probability metrics, genotype likelihood metrics, or genotype quality metrics for genotype calls, based on sequencing metrics extracted from one or more existing sequencing data files. By using a call recalibration machine learning model to update genotype calls based on extracted sequencing metrics, the call recalibration system may improve the accuracy of genotype calls in bi-allelic genomic coordinates, multi-allelic genomic coordinates, or haploid coordinates. The call recalibration system may also recover (i.e., correct) false-negative and false-positive variant calls reported in existing sequencing data files. This disclosure further illustrates such improved accuracy below, at least with respect to Figures 8A and 8B.

[0024] As suggested by the preceding discussion, this disclosure uses various terms to describe the features and advantages of the call recalibration system. Further details regarding the meaning of these terms as used in this disclosure are provided below. As used in this disclosure, for example, the terms “sample nucleotide sequence” or “sample sequence” refer to a sequence of nucleotides isolated or extracted from a sample organism (or a copy of such isolated or extracted sequence). In particular, a sample nucleotide sequence includes a segment of a nucleic acid polymer composed of nitrogen heterocyclic bases, isolated or extracted from a sample organism. For example, a sample nucleotide sequence may include a segment of deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or other polymeric forms of nucleic acids, or chimeric or hybrid forms of nucleic acids as described below. More specifically, in some cases, a sample nucleotide sequence is found in a sample prepared or isolated by a kit and received by a sequencing instrument.

[0025] In relation to this, as used herein, the term “genomic sample” refers to a target genome or portion of a genome that is subjected to assay or sequencing. For example, a genomic sample includes one or more sequences of nucleotides isolated or extracted from a sample organism (or a copy of such isolated or extracted sequences). In particular, a genomic sample includes a whole genome (whole or in part) isolated or extracted from a sample organism and composed of nitrogen heterocyclic bases. For example, a genomic sample may include segments of deoxyribonucleic acid (DNA), ribonucleic acid (RNA), or other polymeric forms of nucleic acids or chimeric or hybrid forms of nucleic acids as described below. In some cases, a genomic sample is one that is prepared or isolated by a kit and found in a sample received by a sequencing instrument.

[0026] Where further used herein, the term “genotype call” refers to the determination or prediction of a particular genotype of a genomic sample at a genomic locus. In particular, a genotype call may include a prediction of a particular genotype of a genomic sample relative to a reference genome or reference sequence at a genomic coordinate or genomic region. For example, in some cases, a genotype call may include a determination or prediction that a genomic sample contains both a nucleobase and a complementary nucleobase at a genomic coordinate that is either homozygous or heterozygous with respect to a reference base or mutation (e.g., a homozygous reference base represented as 0|0, or heterozygous with respect to a particular strand mutation represented as 0|1). Thus, a genotype call may include a prediction of a mutation or reference base for one or more alleles of a genomic sample and may indicate conjugation with respect to the mutation or reference base. Genotype calls are often determined for a genomic coordinate or genomic region where a SNP, insertion, deletion, or other mutation has been identified for a population of organisms.

[0027] Where further used herein, the term “nucleotide base call” (or simply “base call”) refers to the determination or prediction of a specific nucleobase (or nucleobase pair) for an oligonucleotide (e.g., a nucleotide read) during a sequencing cycle, or for genomic coordinates of a sample genome. In particular, a nucleobase call may refer to (i) the determination or prediction of the type of nucleobase embedded within an oligonucleotide on a nucleotide-sample slide (e.g., a read-based nucleobase call), or (ii) the determination or prediction of the type of nucleobase present in genomic coordinates or regions within the genome, including mutational or non-mutational calls in a digital output file. In some cases, for a nucleotide read, a nucleobase call includes the determination or prediction of a nucleobase based on intensity values ​​obtained from fluorescently tagged nucleotides attached to the oligonucleotide on the nucleotide-sample slide (e.g., in a cluster of flow cells). Alternatively, a nucleobase call includes the determination or prediction of a nucleobase from chromatographic peaks or current changes resulting from nucleotides passing through nanopores of a nucleotide-sample slide. In contrast, a nucleobase call may also include a final prediction of the nucleobase in the genomic coordinates of the sample genome for a variant call file (VCF) or another base call output file based on nucleotide reads corresponding to genomic coordinates. Thus, a nucleobase call may include base calls corresponding to genomic coordinates and a reference genome, e.g., an indication of variant or non-variant at a specific location corresponding to the reference genome. In fact, a nucleobase call may refer to a variant call that includes, but is not limited to, a single nucleotide variant (SNV), an insertion or deletion (indel), or a base call that is part of a structural variant. As suggested above, a single nucleobase call may be an adenine (A) call, a cytosine (C) call, a guanine (G) call, a thymine (T) call, or a uracil (U) call.

[0028] In relation to this, as used herein, the term “nucleotide read” refers to the estimated sequence of one or more nucleotide bases (or nucleobase pairs) from all or part of the nucleotide sequence of a sample (e.g., the genomic sequence of the sample, complementary DNA). In particular, a nucleotide read includes the determined or predicted sequence of nucleobase calls for a nucleotide fragment (or group of monoclonal nucleotide fragments) from a sequencing library corresponding to a genomic sample. For example, in some embodiments, a call recalibration system determines the nucleotide read by generating a nucleobase call for nucleobases that have passed through the nanopores of a nucleotide-sample slide, or by determining it via fluorescent tagging, or from wells in a flow cell. In some cases, a nucleotide read may refer to a specific type of read, such as a nucleotide read synthesized from a sample library fragment shorter than a threshold number of nucleobases (e.g., an SBS read). In these or other cases, another type of nucleotide read may refer to (i) a nucleotide read assembled from shorter nucleotide reads to form a continuous sequence (e.g., an assembled nucleotide read) that satisfies the threshold number of nucleobases, (ii) a circular consensus sequencing (CCS) read that satisfies the threshold number of nucleobases, or (iii) a nanopore long read that satisfies the threshold number of nucleobases.

[0029] As described above, in some embodiments, the call recalibration system determines a sequencing metric for nucleotide base calls of nucleotide reads. As used herein, the term “sequencing metric” means a quantitative measurement or score indicating the extent to which an individual nucleotide base call (or sequence of a nucleotide base call) aligns, compares, or quantifies with respect to genomic coordinates or genomic regions of a reference genome, with respect to nucleotide base calls from nucleotide reads, or with respect to external genomic sequencing or genomic structure. For example, a sequencing metric includes a quantitative measurement or score indicating (i) the extent to which an individual nucleotide base call aligns, maps, or covers genomic coordinates or a reference base of a reference genome; (ii) the extent to which a nucleotide base call compares to a reference or alternative nucleotide read with respect to mapping, mismatch, base call quality, or other raw sequencing metrics; or (iii) the extent to which the genomic coordinates or regions corresponding to a nucleotide base call demonstrate mappability, repeat base call content, DNA structure, or other generalized metrics.

[0030] Accordingly, the call recalibration system determines various types of sequencing metrics from different sources, such as read-based sequencing metrics, external source sequencing metrics, and call model-generated sequencing metrics. As used herein, the term “read-based sequencing metric” refers to a sequencing metric derived from nucleotide reads of a sample nucleotide sequence. For example, a read-based sequencing metric includes a sequencing metric determined by applying a statistical test to detect differences between a reference sequence and a nucleotide read. In some embodiments, a read-based sequencing metric may include a comparative mapping quality distribution metric showing a comparison between mapping quality, or a comparative mismatch count metric showing a comparison between mismatch counts. In some cases, a read-based sequencing metric may correspond to genotype calls generated from different read types, such as assembled nucleotide reads and / or SBS reads.

[0031] In contrast, an "externally sourced sequencing metric" refers to a sequencing metric identified or obtained from one or more external databases. For example, an externally sourced sequencing metric may include metrics related to nucleotide mappingability, replication timing, or DNA structure that are available outside of the call recalibration system.

[0032] Furthermore, the term “call-model-generated sequencing metric” refers to an internal model-specific sequencing metric generated or extracted by a call-model-generated model. For example, a call-model-generated sequencing metric may include a variant calling sequencing metric extracted or determined via the variant calling component of the call-model-generated model, and a mapping and alignment sequencing metric extracted or determined via the mapping and alignment component of the call-model-generated model. As shown above, a call-model-generated sequencing metric may include alignment metrics, e.g., deletion size metrics or mapping quality metrics, that quantify the degree to which a nucleotide read aligns with the genomic coordinates of a reference genome or other exemplary nucleic acid sequence. In addition, a call-model-generated sequencing metric may include depth metrics, e.g., forward-reverse-depth metrics or normalized-depth metrics, that quantify the depth of nucleobase calls for a nucleotide read in the genomic coordinates of a reference genome or other exemplary nucleic acid sequence. Call model generation sequencing metrics may also include call quality metrics that quantify the quality or accuracy of nucleobase calls, such as nucleobase-call-quality metrics, callability metrics, or somatic-quality metrics.

[0033] As used herein, the term “base-call-quality metric” refers to a specific score or other measure indicating the accuracy of a nucleo-base call. In particular, a base-call-quality metric includes a value indicating the likelihood that one or more predicted nucleo-base calls to a given genomic coordinate may contain errors. For example, in certain implementations, a base-call-quality metric may include a Q score (e.g., a PHil's Read EDitor (PHRED) quality score) that predicts the probability of error for any given nucleo-base call. For example, a quality score (or Q score) may indicate that the probability of an inaccurate nucleo-base call at a given genomic coordinate is equal to 1 / 100 for a Q20 score, 1 / 1000 for a Q30 score, 1 / 10000 for a Q40 score, and so on.

[0034] In relation to this, in some embodiments, a call recalibration system may generate sequencing metrics by modifying or updating previous metrics. Such “re-engineered sequencing metrics” may refer to sequencing metrics that have been updated, modified, enhanced, refined, or re-engineered to measure or compare nucleobase calls (e.g., nucleobase calls for reads, genotypes, or variant calls) against other nucleobase calls, standards, or references, or targeted for a particular purpose or task. For example, a re-engineered sequencing metric may include modifications to raw (e.g., unmodified) sequencing metrics, or combinations of raw (e.g., unmodified) sequencing metrics. In some embodiments, for example, a call recalibration system generates one or more of the following as re-engineered sequencing metrics: read-based sequencing metrics, external source sequencing metrics, and / or call model generation sequencing metrics. In some cases, remanipulated sequencing metrics refer to sequencing metrics generated by a call recalibration system and therefore residing within or on the call recalibration system itself, and not available to third-party systems. Exemplary remanipulated sequencing metrics include comparative mapping quality distribution metrics that show a comparison between mapping quality distributions associated with a reference sequence and an alternative support nucleotide read, or comparative base quality metrics that show a comparison between base quality of a reference sequence and an alternative support nucleotide read.

[0035] Where used further herein, the term “genomic coordinate” (or sometimes simply “coordinate”) refers to a specific location or position of a nucleobase within a genome (e.g., the genome of an organism or a reference genome). In some cases, genomic coordinates include identifiers for a particular chromosome in the genome and identifiers for the location of a nucleobase within a particular chromosome. For example, genomic coordinates (singular or plural) may include a specific location(singular or plural), such as a number, name, or other identifier for a chromosome (e.g., chr1 or chrX), and a numbered position following the identifier for a chromosome (e.g., chr1:1234570 or chr1:1234570~1234870). In some cases, genomic coordinates refer to genomic coordinates on a sex chromosome (e.g., chrX or chrY). As a result, a call recalibration system may determine genotype probabilities and / or variant call classifications for genotype calls (e.g., variant calls) relating to genomic coordinates on sex chromosomes. Furthermore, in certain implementations, genomic coordinates refer to the source of the reference genome (e.g., mt for the mitochondrial DNA reference genome, or SARS-CoV-2 for the SARS-CoV-2 virus reference genome) and the location of a nucleobase within the source for the reference genome (e.g., mt:16568 or SARS-CoV-2:29001). In contrast, in certain cases, genomic coordinates refer to the location of a nucleobase within the reference genome without referring to a chromosome or source (e.g., 29727).

[0036] In relation to this, as used herein, the term “multi-allele genomic coordinate” refers to a genomic coordinate associated with three or more alleles. For example, a multi-allele genomic coordinate includes a genomic coordinate of nucleotide sequences, where a nucleotide read indicates three or more possible alleles corresponding to the coordinate (e.g., a reference allele, a first alternative allele, a second alternative allele, etc.). In some cases, a multi-allele genomic coordinate corresponds to a genomic coordinate where a read pileup or insertion occurs. For example, a multi-allele genomic coordinate may indicate a multi-allele genotype (e.g., a 1 / 2 genotype), where the first allele in the coordinate corresponds to an allele derived from a first alternative nucleotide sequence, and the second allele corresponds to an allele derived from a second alternative nucleotide sequence.

[0037] As shown above, genomic coordinates within a nucleotide sequence can indicate different genotypes. For example, a “homozygous reference genotype” refers to a genotype (represented as 0 / 0) where both nucleotide bases at a given coordinate in the sample nucleotide sequence match the reference nucleotide base of the reference sequence or reference genome. Another example is a “homozygous surrogate genotype” where both nucleotide bases at a given coordinate differ from the reference nucleotide base of the reference sequence or reference genome (represented as 1 / 1). A further example is a “heterozygous genotype” where the nucleotide bases at a given coordinate are not the same. In some cases, heterozygous genotypes include genotypes (represented as 0 / 1 or 1 / 0) where one nucleotide base matches the reference nucleotide base and the other nucleotide base differs from the reference nucleotide base. With respect to multiple allelic genomic coordinates, a genotype may indicate nucleotide bases derived from two or more surrogate nucleotide bases that differ from the reference nucleotide base of the reference genome. For example, a multiple allele heterozygous genotype can be represented as 1 / 2, where one nucleotide base call matches a first alternative nucleotide base different from the reference nucleotide base, and the other nucleotide base call matches a second alternative nucleotide base different from the reference nucleotide base.

[0038] As described above, genomic coordinates include a location within a reference genome. Such a location may be within a particular reference genome. As used herein, the term “reference genome” refers to a digital nucleic acid sequence assembled as a representative example (or a set of representative examples) of the genes and other gene sequences of an organism. Regardless of sequence length, in some cases a reference genome represents an exemplary set of genes or a set of nucleic acid sequences in a digital nucleic acid sequence determined by scientists as representative of an organism of a particular species. For example, a linear human reference genome may be GRCh38 or other versions of the reference genome from the Genome Reference Consortium. As a further example, a reference genome may include a reference graph genome, e.g., Illumina DRAGEN Graph Reference Genome hg19, which includes both a linear reference genome and paths representing nucleic acid sequences from an ancestral haplotype.

[0039] As suggested above, call recalibration systems can utilize machine learning models to correct sequencing metrics and update genotype calls. As used herein, the term “machine learning model” refers to a computer algorithm or set of computer algorithms that automatically improves for a particular task through experience based on the use of data. For example, a machine learning model may utilize one or more learning techniques to improve accuracy and / or effectiveness. Exemplary machine learning models include various types of decision trees, support vector machines, Bayesian networks, or neural networks. In some cases, a call recalibration machine learning model is a set of gradient-boosted decision trees (e.g., the XGBoost algorithm), and in other cases, a call recalibration machine learning model is a random forest model, a multilayer perceptron, linear regression, a support vector machine, a deep table learning architecture, a deep learning transformer (e.g., a self-attention-based table transformer), or logistic regression.

[0040] In some cases, a call recalibration system utilizes a call recalibration machine learning model to generate an output for verifying, correcting, or updating genotype calls based on extracted sequencing metrics. As used herein, the term “call recalibration machine learning model” refers to a machine learning model that generates a variant call classification. For example, in some cases, a call recalibration machine learning model is trained to generate a variant call classification that shows various probabilities or predictions of genotype calls based on extracted sequencing metrics. Thus, in some cases, a call recalibration machine learning model is a variant call recalibration machine learning model. In some cases, a call recalibration machine learning model is a set of gradient-boosting decision trees (e.g., the XGBoost algorithm, or the treelite algorithm for an ensemble of decision trees), and in other cases, a call recalibration machine learning model is a random forest model, a multilayer perceptron, linear regression, a support vector machine, a deep table learning architecture, a deep learning transformer (e.g., a self-attention-based table transformer), or a logistic regression. In certain embodiments, a call recalibration machine learning model may include multiple submodels or operate in conjunction with another call recalibration machine learning model. For example, a first call recalibration machine learning model (e.g., an ensemble of gradient boosted trees) generates a first variant call classification set, and a second call recalibration machine learning model (e.g., a random forest) generates a second variant call classification set.

[0041] In relation to this, the term “mutation call classification” refers to a predictive classification from a call recalibration machine learning model that exhibits probabilities, scores, or other quantitative measures associated with several aspects of genotype calls based on one or more sequencing metrics extracted from one or more sequencing data files. Mutation call classification may include specific predictions depending on the application of the call recalibration machine learning model. In some cases, mutation call classification for biallelegenic genomic coordinates includes (i) the false positive probability that the genotype call is false positive, (ii) the genotype error probability that the genotype of the genotype call is inaccurate, and (iii) the true positive probability that the genotype call is true positive. In a further embodiment, in an embodiment for generating genotype calls for multiallelegenic genomic coordinates, mutation call classification may include (i) the reference probability that the genotype call contains a homozygous reference genotype in the multiallelegenic genomic coordinates, (ii) the zygosity error probability that the genotype call contains a zygosity error in the multiallelegenic genomic coordinates, and (iii) the true positive variant probability that the genotype call constitutes a true positive variant in the multiallelegenic genomic coordinates.

[0042] In embodiments for generating genotype calls for haploid genomic coordinates, the variant call classification may include (i) a first genotype probability for a first genotype in genomic coordinates and (ii) a second genotype probability for a second genotype in genomic coordinates. As suggested above, the first genotype probability may be the probability that the genotype in genomic coordinates is a haploid reference genotype, and the second genotype probability may be the probability that the genotype in genomic coordinates is a haploid surrogate genotype. In these or other embodiments, for example, embodiments for generating genotype calls for genomic coordinates indicated to represent a homozygous reference genotype, the mutation call classification may include (i) a false positive probability or homozygous reference classification indicating the probability that each genotype call is a false positive or a homozygous reference genotype, (ii) a zygosity error probability or heterozygous genotype classification indicating the probability that each genotype (e.g., an indicator of heterozygous or homozygous genotype for a mutation call at a particular location) is an inaccurate genotype or a heterozygous genotype, and / or (iii) a true positive classification or homozygous alternative classification indicating the probability that each genotype call is a true positive or a homozygous alternative genotype. Thus, in some cases, the mutation call classification represents an intermediate scoring metric and / or a predicted probability that the genotype for a genotype call is accurate.

[0043] Therefore, where used further herein, the term “genotype probability” refers to the likelihood, probability, or score of a particular genotype in a genomic coordinate or genomic region. For example, genotype probability includes the likelihood of a homozygous reference genotype, the likelihood of a heterozygous mutant genotype, or the likelihood of a homozygous mutant genotype in one or more genomic coordinates. In some cases, genotype probability may refer to posterior genotype probability. Therefore, in some cases, genotype probability determined by a call recalibration machine learning model may be presented (or modified to be presented) in the posterior genotype probability (GP) field of a VCF or other sequencing data file, such as a recalibrated VCF or other recalibrated sequencing data file. Genotype probability may include specific predictions depending on the application of the call recalibration machine learning model, such as predicting SNPs.

[0044] As mentioned, in some embodiments, the call recalibration machine learning model can be a neural network. The term “neural network” refers to a machine learning model that can be trained and / or tuned based on inputs for determining classification or approximate unknown functions. For example, a neural network includes a model of interconnected artificial neurons (e.g., organized into layers) that learn to communicate, approximate complex functions, and produce outputs (e.g., a generated digital image) based on multiple inputs given to the neural network. In some cases, a neural network refers to an algorithm (or set of algorithms) that implements deep learning techniques to model high-level abstractions in data. For example, a neural network may include a convolutional neural network, a recurrent neural network (e.g., LSTM), a graph neural network, a self-attentional transform neural network, or a generative adversarial neural network.

[0045] As described above, the call recalibration system can generate a variant call classification that indicates or reflects the likelihood of identifying a variant in genomic coordinates. As used herein, the term “variant” refers to a different or distinct nucleobase or nucleobase that does not align with the corresponding nucleobase(s) in the reference sequence or reference genome. For example, a variant includes a SNP, an indel, or a structural variant that shows a nucleobase in the sample's nucleotide sequence that is different from the nucleobase in the corresponding genomic coordinates of the reference sequence. Accordingly, a “variant nucleotide base call” (or simply “variant call”) refers to a nucleotide base call that contains a variant in a particular genomic coordinate. Conversely, a “non-variant nucleotide base call” (or simply “non-variant call”) refers to a nucleotide base call that contains no variants in genomic coordinates.

[0046] As described above, in one or more embodiments, the call recalibration system extracts or determines sequencing metrics from one or more existing sequencing data files. As used herein, the term “sequencing data file” refers to a digital file containing gene sequencing information relating to genotype calls or nucleotide reads generated by one or more genome sequencing procedures. Such sequencing information may include, for example, nucleotide reads, alignment and mapping information, nucleotide reads at one or more genomic coordinates, etc. In some embodiments, the call recalibration system accesses multiple sequencing data files, such as alignment data files and genotype call data files, to extract or determine sequencing metrics. In one or more embodiments, the call recalibration system accesses a sequencing data file containing all of the aforementioned sequencing information integrated into a single file.

[0047] As mentioned, in some embodiments, the call recalibration system modifies data fields corresponding to genotype call data files, such as variant call files. As used herein, the term “genotype call data file” refers to a digital file that shows or represents one or more genotype calls (e.g., including reference and / or variant calls) compared to a reference genome, along with other information about genotype calls (e.g., variant calls). For example, a genotype call data file may include variant call files such as, but not limited to, variant call format (VCF) files (and genomic variant call format (gVCF) files). Alternatively, in further embodiments, a genotype call data file may include General Feature Format (GFF), Genome Variant Format (GVF), or other suitable data files containing genotype calls for sample nucleotide sequences.

[0048] Where further used herein, “mutation call file” refers to a specific genotype call data file that includes a text file format containing information about mutations at specific genomic coordinates. For example, a mutation call file may include a metadata row, a header row, and data rows, each containing information about a single genotype call (e.g., a single mutation). As further described below, a call recalibration system may generate different versions of genotype call data files, including a pre-filtered mutation call file containing mutation genotype calls that pass or do not pass the quality filter for the base call quality metric, or a post-filtered mutation call file containing mutation genotype calls that pass the quality filter but exclude mutation genotype calls that do not pass the quality filter.

[0049] As also stated, in one or more embodiments, the one or more sequencing data files from which the call recalibration system extracts or determines sequencing metrics include alignment data files containing information from read processing and mapping procedures. As used herein, the term “alignment data file” refers to a digital file that shows mapping and alignment information for nucleotide reads of a sample nucleotide sequence. For example, an alignment data file may include a binary alignment map (BAM) file, a compressed reference-oriented alignment map (CRAM) file, or another file showing nucleotide reads of a sample nucleotide sequence.

[0050] In some embodiments, the call recalibration system modifies data fields corresponding to genotype call metrics associated with a variant call file, such as fields for call quality, genotype, and genotype quality. As used herein, the term “call quality,” when used in relation to data fields within a variant call file, refers to a measure or index of the likelihood or probability that a variant exists at a given location. Thus, a call quality field (or QUAL field) corresponding to a VCF file may include a base call quality metric, such as a PHRED-scaled quality or Q score, representing the probability that the genomic coordinates of the sample genome contain the variant. Similarly, “genotype quality,” when used in relation to a field, refers to the likelihood or probability that a particular predicted genotype for a nucleobase call is accurate.

[0051] As described above, in some embodiments, the call recalibration system extracts or determines sequencing metrics from one or more sequencing data files, such as a genotype call data file containing genotype calls output by a call generation model. As used herein, the term “call generation model” refers to a probabilistic model that generates sequencing data from nucleotide reads of a sample nucleotide sequence, including nucleobase calls, mutation calls, and / or genotype calls, along with the relevant metrics. Thus, in some cases, a call generation model may be a mutation call generation model. For example, in some cases, a call generation model refers to a Bayesian probabilistic model that generates mutation calls based on nucleotide reads of a sample nucleotide sequence. Such a model can process or analyze sequencing metrics corresponding to read pileups (e.g., multiple nucleotide reads corresponding to a single genomic coordinate), which include various hypotheses, including mapping quality, base quality, and foreign read, missing read, and joint detection. A call generation model may also include multiple components, including, but are not limited to, different software applications or components for mapping and alignment, sorting, duplicate marking, calculation of read pileup depth, and mutation calling. In some cases, the call generation model refers to the ILLUMINA DRAGEN model for mutation calling, mapping, and alignment functions (e.g., DRAGEN mutation cola or "DRAGEN VC").

[0052] The following paragraphs describe the call recalibration system with reference to exemplary embodiments and exemplary diagrams illustrating implementation configurations. For example, Figure 1 shows a schematic diagram of a system environment (or "environment") 100 in which the call recalibration system 106 operates, according to one or more embodiments. As shown, the environment 100 includes one or more server devices 102 connected to client devices 108 and sequence determination devices 114 via a network 112. While Figure 1 shows one embodiment of the call recalibration system 106, alternative embodiments and configurations are described below.

[0053] As shown in Figure 1, the server device 102, the client device 108, and the sequence determination device 114 can communicate with each other via the network 112. The network 112 includes any suitable network on which the computing devices can communicate. An exemplary network will be discussed in more detail below with respect to Figure 10.

[0054] As shown in Figure 1, the sequencing device 114 includes an apparatus for sequencing nucleic acid polymers. In some embodiments, the sequencing device 114 analyzes nucleic acid segments or oligonucleotides extracted from a genomic sample to generate nucleotide reads or other data, either directly or indirectly on the sequencing device 114 using computer methods and systems (as described herein). More specifically, the sequencing device 114 receives and analyzes nucleic acid sequences extracted from a genomic sample in a nucleotide-sample slide. In one or more embodiments, the sequencing device 114 sequences nucleic acid polymers into nucleotide reads using SBS. In some embodiments, in addition to communicating via the network 112, or as an alternative thereto, the sequencing device 114 bypasses the network 112 and communicates directly with the client device 108.

[0055] As further shown in Figure 1, the server device 102 can generate, receive, analyze, store, and transmit digital data such as data for determining nucleotide-based calls or sequencing nucleic acid polymers. As shown in Figure 1, the sequencing device 114 can transmit call data from the sequencing device 114 (and the server device 102 can receive call data). The server device 102 can also communicate with the client device 108. In particular, the server device 102 can transmit data to the client device 108 that includes sequencing data files such as genotype call data files or alignment data files, or other information indicating nucleotide-based calls, sequencing metrics, error data, or other metrics associated with nucleotide-based calls or genotype calls.

[0056] In some embodiments, the server device 102 comprises a distributed collection of servers, and the server device 102 includes several server devices distributed across the network 112 and located in the same or different physical locations. Furthermore, the server device 102 may include content servers, application servers, communication servers, web hosting servers, or other types of servers. In some cases, the server device 102 is located in the same physical location as the sequencing device 114.

[0057] As further shown in Figure 1, the server device 102 may include a sequencing system 104. Generally, the sequencing system 104 analyzes call data, such as sequencing data received from the sequencing device 114, to determine nucleotide-based sequences for nucleic acid polymers. For example, the sequencing system 104 may receive raw data from the sequencing device 114 and determine nucleotide-based sequences for sample nucleotide sequences (e.g., genomic samples). In some embodiments, the sequencing system 104 determines nucleotide-based sequences in DNA and / or RNA segments or oligonucleotides. In addition to processing and determining sequences for nucleic acid polymers, the sequencing system 104 also generates genotype call data files, such as a mutation call file, which shows one or more genotype calls and / or mutation calls for one or more genomic coordinates.

[0058] As described above and as shown in Figure 1, the call recalibration system 106 analyzes call data, such as sequencing metrics from the sequencing instrument 114 stored in an existing sequencing data file, to recalibrate genotype calls for previously generated sample nucleotide sequences (e.g., by a call generation model). The call recalibration system 106 includes a call recalibration machine learning model. In some embodiments, the call recalibration system 106 determines sequencing metrics for sample nucleotide sequences based on information stored in an existing sequencing data file. Based on data derived or prepared from the sequencing metrics, the call recalibration system 106 trains and applies a call recalibration machine learning model to recalibrate genotype calls for sample sequences corresponding to genomic coordinates. The call recalibration system 106 further utilizes the call recalibration machine learning model to generate a set of variant call classifications to update or correct genotype calls (e.g., variant calls). Based on such data, for example, the call recalibration system 106 may update the genotype calls (e.g., mutation calls) to improve accuracy by updating data fields corresponding to genotype call files, such as mutation call files. In some embodiments, the call recalibration system 106 outputs an updated mutation call file (or a genotype call data file in another format) having the modified or updated genotype calls and / or mutation calls.

[0059] As further illustrated in Figure 1, the client device 108 can generate, store, receive, and transmit digital data. In particular, the client device 108 can receive sequencing metrics from the sequencing device 114. Furthermore, the client device 108 can communicate with the server device 102 to receive genotype call data files, such as genotype call files and / or mutation call files containing other metrics such as call quality, genotype index, and genotype quality. Thus, the client device 108 can present or display information about genotype calls within a graphical user interface to the user associated with the client device 108.

[0060] The client device 108 illustrated in Figure 1 can include various types of client devices. For example, in some embodiments, the client device 108 includes non-mobile devices such as desktop computers or servers, or other types of client devices. In yet other embodiments, the client device 108 includes mobile devices such as laptops, tablets, mobile phones, or smartphones. Further details regarding the client device 108 are discussed below with respect to Figure 10.

[0061] As further illustrated in Figure 1, the client device 108 includes a sequencing application 110. The sequencing application 110 may be a web application or a native application (e.g., a mobile application, a desktop application) stored and executed on the client device 108. The sequencing application 110 may include instructions to cause the client device 108 to receive data from the call recalibration system 106 and to present data from the variant call file and / or updated variant call file for display on the client device 108. Furthermore, the sequencing application 110 may instruct the client device 108 to display a visualization of sequencing metrics for nucleotide-based calls or genotype calls.

[0062] As further illustrated in Figure 1, the call recalibration system 106 may be located on the client device 108 or on the sequencing device 114 as part of the sequencing application 110. Therefore, in some embodiments, the call recalibration system 106 is implemented on the client device 108 (e.g., entirely or partially located). In yet other embodiments, the call recalibration system 106 is implemented by one or more other components of the environment 100, such as the sequencing device 114. In particular, the call recalibration system 106 can be implemented in various different ways across the server device 102, the network 112, the client device 108, and the sequencing device 114. For example, the call recalibration system 106 can be downloaded from the server device 102 to the client device 108 and / or the sequencing device 114, and all or part of the functions of the call recalibration system 106 are performed on each device within the environment 100.

[0063] As further illustrated in Figure 1, the environment 100 includes a database 116. The database 116 may store information such as sequencing data files 118, sample nucleotide sequences, nucleotide reads, nucleotide base calls, genotype calls (e.g., mutation calls), and sequencing metrics. In some embodiments, a server device 102, a client device 108, and / or a sequencing device 114 communicate with the database 116 (e.g., via a network 112) to store and / or access information such as sequencing data files 118, sample nucleotide sequences, nucleotide reads, nucleotide base calls, genotype calls (e.g., mutation calls), and sequencing metrics. In some cases, the database 116 also stores one or more models, such as call recalibration machine learning models.

[0064] Figure 1 illustrates components of environment 100 communicating via network 112; however, in certain implementations, components of environment 100 may also communicate directly with each other, bypassing network 112. For example, as previously mentioned, in some implementations, client device 108 may communicate directly with sequence determination device 114. Additionally, in some embodiments, client device 108 communicates directly with call recalibration system 106. Furthermore, call recalibration system 106 may access one or more databases housed in or accessed by server device 102 or other locations within environment 100.

[0065] As described above, the call recalibration system 106 can generate modified or updated genotype calls based on information extracted or determined from one or more existing sequencing data files. As also mentioned, in some implementations, the call recalibration system 106 exhibits improved efficiency and / or accuracy compared to existing sequencing systems when generating recalibrated and / or improved genotype calls for previously analyzed sample nucleotide sequences.

[0066] To illustrate further, Figure 2A outlines an existing sequencing system that generates updated sequencing data for a previously analyzed sample nucleotide sequencing system. Specifically, Figure 2A shows two existing call generation models, call generation model 204a and updated call generation model 204b, which generate genotype calls from base call data 202 and output each genotype call to the respective sequencing data file 210a and updated sequencing data file 210b.

[0067] As shown in the examples provided, existing methods for updating genotype calls and other sequencing information may include utilizing updated call generation models (e.g., newer call generation models showing improved results), such as updated call generation model 204b, to reanalyze data from sequencing instruments, such as base call data 202. As suggested above, in some cases, updated call generation model 204b includes updated versions of machine learning-based mutant calls that are lacking in call generation model 204a. In some cases, a large number of sequencing data files (e.g., sequencing data files containing nucleotide base reads for thousands of sample nucleotide sequences) are updated in this manner, requiring extensive computational resources to reprocess the base call data for each of the previously analyzed genomic samples.

[0068] For example, as shown, the analysis of base call data 202 by the call generation model 204a includes a procedure for read processing and mapping 206a and a procedure for genotype calling 208a. When newly analyzed by the updated call generation model 204b, the updated analysis generally requires the execution of updated read processing and mapping procedure 206b and updated genotype calling procedure 208b.

[0069] In contrast, as shown in Figure 2B, the call recalibration system 106 verifies, updates, or corrects genotype calls based on information extracted or determined from an existing sequencing data file 210a previously generated by the call generation model 204a. In other words, the call recalibration system 106 generates a recalibrated sequencing data file 210c without directly reprocessing the base call data 202. Instead, as shown, the call recalibration system 106 utilizes the call recalibration machine learning model 212 to generate a recalibrated sequencing data file 210c based on sequencing metrics extracted or determined from the existing data file 210a. Thus, the call recalibration system 106 leverages the previous analysis of the base call data 202 by the call generation model 204a, including read processing and mapping 206a and genotype calls 208a, by utilizing the trained call recalibration machine learning model 212 to update the sequencing data in the existing sequencing data file 210a. As shown in Figure 7 and other figures, the call recalibration system 106 saves considerable computer processing runtime and extends the flexibility of the processor type for variant calls by generally following the procedure in Figure 2B rather than the procedure in Figure 2A.

[0070] As described above, the call recalibration system 106 can generate mutation call classifications based on sequencing metrics extracted or determined from one or more existing sequencing data files. In particular, the call recalibration system 106 can determine mutation call classifications from extracted sequencing metrics using a call recalibration machine learning model, and can determine or update various metrics associated with genotype calls from the generated mutation call classifications. Figure 3 shows an exemplary overview of the call recalibration system 106, which determines mutation call classifications based on extracted sequencing metrics, according to one or more embodiments.

[0071] As shown in Figure 3, the call recalibration system 106 performs operation 302 to receive one or more existing sequencing data files. In particular, in some embodiments, the call recalibration system 106 receives a first sequencing data file, such as an alignment data file (e.g., a BAM file or CRAM file) containing nucleotide read data. In addition, in some embodiments, the call recalibration system 106 receives a second sequencing data file, such as a genotype call data file (e.g., a VCF file or gVCF file) having one or more genotype calls at one or more genomic coordinates. Alternatively, in one or more embodiments, the call recalibration system 106 may receive the aforementioned sequencing data in one or more sequencing data files of an alternative format, such as, but not limited to, a single sequencing data file having data about nucleotide reads and genotype calls. Furthermore, in some embodiments, the call recalibration system 106 receives one or more sequencing data files having additional sequencing information, which will be discussed further below in relation to subsequent figures.

[0072] As shown in Figure 3, the call recalibration system 106 performs operation 304 to extract sequencing metrics from one or more sequencing data files received during operation 302. In particular, the call recalibration system 106 extracts or determines sequencing metrics such as read-based sequencing metrics and call-based sequencing metrics. For example, the call recalibration system 106 extracts or determines sequencing metrics that show various attributes or data regarding various genotype calls from a sample nucleotide sequence (e.g., a genomic sample). In some embodiments, the call recalibration system 106 determines or extracts different sequencing metrics to generate genotype calls associated with different variants such as SNPs and indels. Furthermore, in some embodiments, as shown in Figure 3, the call recalibration system 106 performs operation 306 to access or receive sequencing metrics from an external source. Further details regarding the determination of various types of sequencing metrics are provided below with reference to Figures 4A–4C.

[0073] As further illustrated in Figure 3, the call recalibration system 106 performs operation 308 to generate a variant call classification. More specifically, the call recalibration system 106 uses a call recalibration machine learning model to generate (or update or improve) a variant call classification from extracted sequencing metrics. In detail, in some embodiments, the call recalibration system 106 uses a call recalibration machine learning model to process or analyze one or more extracted sequencing metrics to generate a set of classifications (e.g., predicted probabilities associated with genotypes). For example, the call recalibration system 106 uses a call recalibration machine learning model to generate a set of variant call classifications (represented in Figure 3 as "Class 1", "Class 2", and "Class 3") that indicate a specific probability associated with a genotype for a corresponding genotype call, based on the sequencing metrics.

[0074] In some embodiments, the call recalibration system 106 generates different variant call classifications for different applications and / or different genomic coordinates. By using different versions of the call recalibration machine learning model, for example, the call recalibration system 106 generates a first set of variant call classifications for multi-allele genomic coordinates, a second set of variant call classifications for haploid genomic coordinates, and a third set of variant call classifications for genomic coordinates where homozygous reference genotypes are indicated. In certain embodiments, the call recalibration system 106 generates the same variant call classifications for different applications and / or different genomic coordinates, but they are used separately or different information associated with the variant call classifications is used. Further details regarding the generation of variant call classifications are provided below with reference to the following figures.

[0075] As further illustrated in Figure 3, the call recalibration system 106 also performs operation 310, which generates one or more recalibrated sequencing data files. For example, the call recalibration system 106 determines a modified or updated genotype call (e.g., a variant call) based on a variant call classification and indicates any such modifications in the recalibrated sequencing data file (e.g., an updated VCF file). More specifically, the call recalibration system 106 modifies or updates the genotype call for the sample nucleotide sequence at genomic coordinates within the reference genome. To determine the updated or modified genotype call, in some embodiments, the call recalibration system 106 edits or updates a specific existing genotype call (i.e., from the sequencing data file received in operation 302) based on a variant call classification generated by a call recalibration machine learning model.

[0076] More specifically, the call recalibration system 106 extracts or determines sequencing metrics (e.g., one or more of the same sequencing metrics used to generate variant call classifications) and analyzes genotype calls from the extracted sequencing metrics. For example, the call recalibration system 106 applies several Bayesian probabilistic models or algorithms to derive various probabilities for different nucleotide bases, quality metrics, mapping metrics, joint metrics, and other data occurring within the sample nucleotide sequence. From the probabilistic models, the call recalibration system 106 determines updated genotype calls (e.g., calls indicating a difference or identity to a reference base from a reference genome) that represent a pair of predicted nucleotide bases of the genomic sample at corresponding genomic coordinates.

[0077] As further illustrated in Figure 3, in a particular implementation, the call recalibration system 106 utilizes variant call classification (such as that determined via operation 308) to generate, recalibrate, modify, or enhance existing genotype calls. More specifically, the call recalibration system 106 utilizes probabilities associated with variant call classification to determine or update certain metrics associated with genotype calls. For example, the call recalibration system 106 modifies data fields corresponding to sequencing data files (e.g., genotype call data file, variant call file, etc.) for metrics such as call quality, genotype, and genotype quality (or others described below).

[0078] In some cases, the call recalibration system 106 extrapolates from the variant call classification to determine metrics corresponding to existing sequencing data files, such as call quality, genotype, and genotype quality associated with genotype calls. For example, by utilizing the zygosity error probability, the call recalibration system 106 may repair certain errors in or associated with existing genotype calls. In fact, if the call recalibration system 106 determines a high false-positive probability for a genotype call, the call recalibration system 106 applies a call recalibration machine learning model to act as a variant filter to correct (e.g., reduce) the call quality associated with the genotype call. In another embodiment, if the system previously filtered out or double-penalized heterozygous / homozygous (het / hom) errors (e.g., if the system generates an incorrect genotype call, causing it to miss a more accurate genotype call), the call recalibration system 106 utilizes the zygosity error probability to correct the genotype and / or genotype quality of the genotype call.

[0079] In relation to this, in some embodiments, the call recalibration system 106 generates updated genotype calls in biallele (or other types of) genomic coordinates by deciding to change existing genotype calls from homozygous to heterozygous, from heterozygous to homozygous, from mutant calls to reference calls, from reference calls to mutant calls, or any combination of the above. For example, in some implementations, the call recalibration system 106 changes heterozygous mutant genotype calls or homozygous mutant genotype calls reported in one or more sequencing data files to homozygous reference genotype calls in genomic coordinates. In another implementation, the call recalibration system 106 changes homozygous reference genotype calls or homozygous mutant genotype calls reported in one or more sequencing data files to heterozygous mutant genotype calls in genomic coordinates. In yet another implementation, the call recalibration system 106 changes heterozygous mutant genotype calls or homozygous reference genotype calls reported in one or more sequencing data files to homozygous mutant genotype calls in genomic coordinates.

[0080] In certain embodiments, the call recalibration system 106 modifies a data field for genotype calls (e.g., call quality, genotype, or genotype quality) by considering a single variant call classification. In contrast, in some embodiments, the call recalibration system 106 modifies or updates one or more data fields for call quality, genotype, and / or genotype quality by considering multiple variant call classifications at once (e.g., in weighted combinations). Further details regarding the generation and modification of genotype calls are provided below with reference to subsequent figures.

[0081] As described above, in some implementations, the call recalibration system 106 is not necessarily limited to correcting or updating a single genotype call contained in one or more sequencing data files. For example, in one or more embodiments, the call recalibration system 106 extracts sequencing metrics for a particular genotype call from one or more sequencing data files by extracting sequencing metrics for a particular genotype call, generating variant call classifications for a genotype call, and / or correcting the genotype call based on the variant call classifications. By utilizing a call recalibration machine learning model, the call recalibration system 106 generates one or more variant call classifications that indicate the accuracy of a particular genotype call as shown by one or more sequencing data files. Based on one or more additional variant call classifications, the call recalibration system 106, in some implementations, corrects or updates a base call quality metric (e.g., Q score), a genotype probability metric, a genotype likelihood metric, or a genotype quality metric for a particular genotype call. For example, in some implementations, the call recalibration system 106 determines a modified base call quality metric, modified genotype probability metric, modified genotype likelihood metric, or modified genotype quality metric for specific genotype calls that fall below the base call quality threshold or other corresponding thresholds. Based on the fact that the modified base call quality metric does not meet the base call quality threshold, the call recalibration system 106 annotates the specific genotype call in the recalibrated sequencing data file to indicate that the modified base call quality metric falls below the base call quality threshold. Similarly, based on the modified genotype probability metric, modified genotype likelihood metric, or modified genotype quality metric that do not meet the genotype probability threshold, genotype likelihood threshold, or genotype quality threshold, respectively, the call recalibration system 106 annotates the specific genotype call in the recalibrated sequencing data file to indicate that the modified genotype probability metric, modified genotype likelihood metric, or modified genotype quality metric falls below the corresponding threshold.

[0082] As mentioned above, in certain described embodiments, the call recalibration system 106 extracts sequencing metrics for nucleotide-based calls or genotype calls at specific genomic coordinates. In particular, the call recalibration system 106 extracts sequencing metrics such as read-based sequencing metrics, external source sequencing metrics, and call model generation sequencing metrics from one or more sequencing data files for calls corresponding to existing nucleotide reads from a sample nucleotide sequence. Figures 4A to 4C illustrate the extraction of sequencing metrics according to one or more embodiments. Specifically, Figure 4A shows receiving sequencing data files (alignment data file 406 and genotype call data file 410) containing sequencing data for nucleotide reads 402 of a sample nucleotide sequence, Figure 4B shows determining an external source sequencing metric 414, and Figure 4C shows extracting the sequencing metric 416 and generating a recalibrated sequencing data file 422.

[0083] As shown in Figure 4A, the call recalibration system 106 accesses, retrieves, or otherwise obtains nucleotide reads 402. In particular, in some embodiments, the nucleotide reads 402 are pre-determined using the sequencing instrument 114 and include nucleotide base calls for regions from a sample nucleotide sequence (e.g., a sample genome). For example, the nucleotide reads 402 may be generated using synthetic sequencing (SBS) techniques and / or Sanger sequencing techniques to determine nucleotide base calls for oligonucleotide clusters from wells in a flow cell and / or via fluorescent tagging. More specifically, the nucleotide reads 402 are generated using cluster generation and SBS chemistry to sequence millions or billions of clusters in a flow cell. During SBS chemistry, for each cluster, the call nucleotide base calls from the nucleotide reads 402 are stored and, in some embodiments, provided directly to the call recalibration system 106 for each sequencing cycle via real-time analysis (RTA) software.

[0084] As further shown in Figure 4A, in some embodiments, alignment data files 406, such as BAM files or CRAM files, are generated by read processing and mapping 404. For example, read processing and mapping 404 includes using real-time analysis (RTA) software to store base call data in the form of individual base call data files (or BCLs). In some cases, read processing and mapping 404 further includes converting the BCL files into sequence data to be analyzed by a call generation model 408 (e.g., via conversion from BCL to FASTQ) to determine genotype calls to nucleotide reads 402.

[0085] In particular, in certain embodiments, read processing and mapping 404 includes aligning nucleotide reads with a reference genome or receiving information regarding read alignment. Specifically, read processing and mapping 404 determines (or receives information indicating alignment) which nucleotide bases of a given read align with which genomic coordinates of the reference sequence. Different reads have different lengths and contain different nucleotide bases. Therefore, in some cases, read processing and mapping 404 includes analyzing each nucleotide of each read to determine (or receive information indicating) where the read "fits" with respect to the reference sequence, for example, where a base in the read aligns with a base in the reference. In some cases, read processing and mapping 404 includes aligning many reads at a single genomic coordinate, thus resulting in a read pileup.

[0086] In certain embodiments, the call recalibration system 106 performs additional statistical tests to determine or detect differences between metrics associated with a reference nucleotide sequence and metrics associated with alternative support nucleotide reads. Through these statistical tests, the call recalibration system 106 reworks the raw sequencing metrics to determine read-based sequencing metrics. In some cases, the call recalibration system 106 extracts raw sequencing metrics that include one or more of the following: (i) an alignment metric for quantifying the alignment of a sample nucleotide sequence with the genomic coordinates of an exemplary nucleotide sequence (e.g., a nucleotide sequence from a reference genome or ancestral haplotype); (ii) a depth metric for quantifying the depth of the nucleotide-based call for the sample nucleotide sequence at the genomic coordinates of the exemplary nucleotide sequence; or (iii) a call quality metric for quantifying the quality of the nucleotide-based call (e.g., a genotype call) for the sample nucleotide sequence at the genomic coordinates of the exemplary nucleotide sequence. For example, the call recalibration system 106 extracts a mapping quality metric (e.g., the MAPQ metric shown in Figure 4A), a soft clipping metric, or other alignment metrics that measure the alignment of the sample sequence with a reference genome. In another embodiment, the call recalibration system 106 extracts a forward-reverse depth metric (or other such depth metric) or a callability metric (or other such call quality metric) for mutant genotype calls.

[0087] As previously mentioned, in some embodiments, the call recalibration system 106 reworks the raw sequencing metrics extracted from the alignment data file 406 (or other sequencing data file) to generate more useful read-based sequencing metrics for comparing metrics associated with the reference nucleotide sequence with metrics associated with various support alternative nucleotide reads. For example, the call recalibration system 106 extracts various metrics for the sample sequence with respect to the reference sequence, and further extracts various metrics for the sample sequence with respect to the alternative support sequence. In addition, in some embodiments, the call recalibration system 106 performs a comparative analysis between metrics associated with the reference sequence and metrics associated with the alternative support reads.

[0088] For example, the call recalibration system 106 compares how the nucleotide bases of a sample nucleotide sequence (e.g., a sample genome) map to a reference sequence with how the nucleotide bases map to various alternative support reads. In some cases, the call recalibration system 106 determines the mapping quality associated with the reference sequence to compare with the mapping quality associated with the alternative support reads. For example, the call recalibration system 106 determines mapping quality statistics that reflect the difference in the distribution of contrasts between reads supporting the reference sequence and reads supporting alternative alleles.

[0089] In these or other cases, the call recalibration system 106 determines the mismatch count between the sample sequence and the reference sequence, and between the reference sequence and the alternative support read. The call recalibration system 106 further compares the mismatch counts to extract a comparative mismatch count metric. Furthermore, the call recalibration system 106 determines the soft clipping metric for the sample sequence with respect to the reference sequence and further extracts the soft clipping metric for the alternative support read. The call recalibration system 106 also compares the soft clipping metrics between the reference sequence and the alternative support read to generate a comparative soft clipping metric. Furthermore, the call recalibration system 106 compares the base call quality metrics for the reference sequence and the alternative support read, and / or compares the query positions of the sample sequence with respect to the reference sequence and the query positions with respect to the alternative support read.

[0090] As further illustrated in Figures 4A and 4C, the call recalibration system 106 uses comparative and / or other statistical tests to extract read-based sequencing metrics from information in the alignment data file 406 (or other sequencing data file), including: i) a comparative mapping quality distribution metric showing a mapping quality distribution comparing the mapping quality for the reference sequence with the mapping quality for the alternative support reads; ii) a comparative secondary mapping alignment metric showing a comparison between secondary mappings for bases in the reference sequence and secondary mappings for bases in the alternative support reads; iii) a comparative mismatch count metric showing a comparison between mismatch nucleotide bases for the reference sequence and mismatch bases for the alternative support reads; iv) a comparative soft clipping metric showing a comparison between soft clipping metrics for the reference sequence and soft clipping metrics for alternative support reads; v) nucleo vi) one or more read depth comparison metrics showing a comparison between the read depth of rheotide read 402 and one or more average read depths (e.g., local average read depth at a specific genomic coordinate and global average read depth across multiple genomic coordinates within a certain region); vi) one or more comparative base quality metrics showing a comparison between base quality for the reference sequence and base quality for alternative support reads (e.g., overall base quality, early base quality, and late base quality for nucleotide read 402); vii) one or more comparative query location metrics showing a comparison between query location for the reference sequence and query location for alternative support reads; viiii) one or more contextual information metrics showing homopolymerization and periodicity of nucleotide base calls; ix) a strand bias metric showing strand bias associated with one or more nucleotide reads 402; and x) a read direction bias metric showing read direction bias associated with nucleotide read 402. In some cases, the call recalibration system 106 extracts or redesigns additional or alternative read-based sequencing metrics from information stored in the alignment data file 406 (or other sequencing data file).

[0091] In addition to read-based sequencing metrics, as shown in Figures 4A and 4C, the call recalibration system 106 extracts call model-generated sequencing metrics from the genotype call data file 410 generated by the call generation model 408. For example, the call generation model 408 determines sequence data based on read processing and mapping 404. In some cases, the call generation model 408 generates sequence data as part of one or more digital files, such as BCL and FASTQ files.

[0092] To generate such files, in some embodiments, the sequencing device 114 (or call generation model 408) sequences millions or billions of clusters in a flow cell using cluster generation and SBS chemistry. During SBS chemistry, for each cluster, the sequencing device 114 (or call generation model 408) stores the nucleotide base calls from the nucleotide read 402 for each sequencing cycle via real-time analysis (RTA) software. The sequencing device 114 (or call generation model 408) further stores the base call data in the form of individual base call data files (or BCLs) using the RTA software. In some cases, the sequencing device 114 (or call generation model 408) further converts the BCL files into sequence data (e.g., via BCL to FASTQ conversion). For example, the sequencing device 114 (or call generation model 408) generates a FASTQ file from the nucleotide read 402, and the FASTQ file contains the sequence data.

[0093] In some cases, the call generation model 408 generates sequence data for each cluster that passes through the initial quality filter from the sample sequence. For example, the call generation model 408 generates an entry for each cluster, each entry containing the following four rows (or four items of sequence data): i) a sequence identifier with information about the sequencing process and the cluster, ii) the nucleotide base calls that make up the sequence (e.g., the sequences of A, C, T, G, and / or N calls), iii) a separator (e.g., a "+" symbol), and iv) a base call quality metric indicating the probability of accuracy for the nucleotide base calls (Phred+33 coding).

[0094] As further shown in Figure 4A, the call generation model 408 processes or analyzes sequence data to generate genotype calls. In fact, in some embodiments, the call recalibration system 106 extracts call model-generated sequence determination metrics by redesigning raw sequence determination metrics (e.g., raw sequence determination metrics in sequence data utilized by the call generation model 408 and stored in one or more sequence data files such as BCL or FASTQ files). In particular, the call generation model 408 includes a mapping and alignment component for mapping and aligning nucleotide-based calls from sequence data. In addition, the call generation model 408 includes a variant call component that generates genotype calls (e.g., reference-based calls such as variant calls or non-variant calls) from sequence data and stores the genotype calls in a genotype call data file 410 (e.g., a VCF or gVCF file). In some cases, the call recalibration system 106 extracts previously generated call model generation sequencing metrics by utilizing the mapping and alignment components and mutation call components of the call generation model 408 by accessing the genotype call data file 410.

[0095] To illustrate an example of a call model generation sequencing metric, in some cases the call recalibration system 106 may include: i) a base call quality metric (e.g., DRAGEN QUAL score) indicating the quality score for genotype calls generated via the call generation model 408; ii) a call model generation foreign read detection metric (e.g., foreign read detection (FRD) score) indicating the probability that one or more of the nucleotide reads 402 during pile-up may be foreign reads (e.g., their true locations are elsewhere in the reference sequence); iii) a call model generation base quality dropoff metric (e.g., base quality dropoff (BQD) score) indicating the probability of base quality dropoff based on one or more of the following: strand bias, error locations in the thread, or low mean base quality across a subset of nucleotide reads 402; iv) mean read depth; and v) indel statistics (e.g., polymerase chain reaction). Extracting variant calling metrics (including one or more of the following: reaction, PCR curve) and / or vi) Hidden Markov Model (HMM) statistics (or reconstructing Hidden Markov Model (HMM) statistics using a concise and characteristic gapped alignment report (CIGAR)); vii) secondary alignment metric indicating the probability that a secondary genotype call is accurate; viii) base context metric indicating contextual information for nucleotides surrounding a genotype call; iv) neighbor call metric indicating the neighborhood of a genotype call (e.g., adjacent or within a threshold of separation from it); x) joint detection metric indicating the probability of detecting a joint corresponding to two or more duplicate nucleotide base calls; xii) read filtering metric indicating a threshold quality metric or other metric for filtering out nucleotide base calls having low mapping quality, base quality, or other quality metrics, etc.In some cases, the call recalibration system 106 extracts the call model-generated sequence decision metric from internal (e.g., unique and model-specific) variables that reflect interacting processing paths, corner cases, and difficult predictions / decisions.

[0096] Although not listed in Figure 4A or Figure 4C, the call model generation sequence determination metrics include, but are not limited to, the variant call metrics extracted via the variant call component of the call generation model 408 and stored in (or otherwise determined from) the existing version of the genotype call data file 410. In addition to, or as an alternative to, the call recalibration system 106 may, in some cases, use the following criteria: i) the number of samples in the population; ii) the number of reads processed to generate genotype calls, the number of mutations (e.g., SNPs, indels, and MNPs); iii) the number of bialleletic sites (e.g., genomic coordinates containing two observed alleles); iv) the number of multialleletic sites (e.g., the number of sites in the mutation call file containing three or more observed alleles); v) the number of SNPs; vi) the number of different types of indels (e.g., homozygous insertions, heterozygous insertions, and heterozygous deletions); vii) the total number of heterozygous indels (e.g., insertion + deletion, insertion + SNP, or deletion + SNP); viiii) the number of de novo SNPs (e.g., de novo SNPs that meet a threshold level). Extract or generate a mutation calling metric (e.g., via metric remanipulation) that includes one or more of the following: x) SNPs with a novo quality metric, x) the number of de novo indels (e.g., SNPs with a novo quality metric that meets a threshold level), x) the number of de novo MNPs (e.g., MNPs with a novo quality metric that meets a threshold level), x) the number of SNPs in chromosome 1 divided by the number of SNPs in chromosome 2, xii) the number of SNP base transpositions, xiii) the number of SNP transversions, xiv) the number of heterozygous mutations, xv) the number of homozygous mutations, xvi) the ratio between the number of heterozygous mutations and the number of homozygous mutations, xvii) the number of mutations detected in the dbSNP reference file, and / or xviii) the total number of mutations minus the number detected in the dbSNP file.

[0097] Additionally, the call model generation sequencing metrics may include mapping and alignment sequencing metrics extracted via the mapping and alignment components of the call generation model 408 and stored in existing versions of the genotype call data file 410 and / or alignment data file 406. For example, the call recalibration system 106 may include i) the total number of input reads, ii) the number of duplicate-marked reads, iii) the number of mate reads with duplicate marks removed, iv) the number of unique reads, v) the number of reads with mate sequences, vi) the number of reads without mate sequences, vii) an index of reads that fail quality checks, viiii) an index of mapped reads, ix) the number of unique and mapped reads, x) the number of unmapped reads, xi) singleton reads. xii) number of reads (e.g., reads are mapped but their pairmates are not), xiii) number of paired reads, xiv) number of properly paired reads (e.g., both reads in a pair are mapped and fall within acceptable limits based on the estimated insert length distribution), xiv) number of mismatched reads (e.g., the number of reads that are not properly paired), xv) number of paired reads mapped to different chromosomes, xvi) number of paired reads mapped to different chromosomes and with a mapping quality metric of 10 or higher, xvii) indel R1 and Mapping and alignment metrics are extracted or generated (e.g., via metric remanufacturing) including one or more of the following: xviii) the percentage of reads in R2, xix) the percentage of soft-clipped bases in R1 and R2, xix) the number of mismatched bases in R1 and R2, xx) the number of bases having at least 30 base quality (e.g., total and / or R1 or R2), xxi) the number of alignments (e.g., total alignment, secondary alignment, and / or supplemental alignment), xxii) estimated read length, and xxiii) estimated sample contamination.

[0098] Referring here to Figure 4B, the call recalibration system 106 may, in some cases, generate, extract, or determine an external source sequencing metric 414. In particular, the call recalibration system 106 determines an external source sequencing metric 414 from one or more external databases, for example, a sequencing information database 412 (e.g., database 116). For example, the call recalibration system 106 accesses a common or generally applicable sequencing metric for sequencing nucleotides. In addition, the call recalibration system 106 accesses or determines sequencing information (e.g., stored in the sequencing information database 412) for a specific reference sequence. In some cases, the call recalibration system 106 determines sequencing metrics 414 of an external source, including: i) a mapping feasibility metric indicating the ease or difficulty of mapping a particular nucleotide sequence (or a particular nucleotide read or nucleotide base call); ii) a guanine-cytosine content metric indicating the count (or dropout or mean) of guanine-cytosine content in a reference nucleotide sequence (e.g., a reference genome); iii) a replication timing metric indicating the time required to replicate a specific number of nucleotides from the reference sequence; iv) one or more DNA structure metrics indicating the DNA structure of the reference sequence (e.g., a reference genome); v) a conservation metric indicating a measure of sequence conservation across multiple species (e.g., a measure of change relative to the mean); and / or others.

[0099] As discussed above (for example, in relation to Figure 3), and as shown in Figure 4C, the call recalibration system 106 extracts sequencing metrics 416 from sequencing data files previously generated by various sequencing and base calling processes, such as read processing and mapping 404 and base calling by the call generation model 408 (see Figure 4A). In particular, the call recalibration system 106 uses the call recalibration machine learning model 418 to generate a variant call classification 420, and based on the variant call classification 420, generates a recalibrated sequencing data file 422 as output.

[0100] In one or more embodiments, the call recalibration system 106 extracts (or reconstructs) additional or alternative sequencing metrics from one or more sequencing data files, including read-based sequencing metrics, call model-generated sequencing metrics, and / or external source sequencing metrics. For example, the call recalibration system 106 extracts sequencing metrics in the following table, where each metric belongs to one or more of the read-based sequencing metrics, call model-generated sequencing metrics, and / or external source sequencing metrics.

[0101] [Table 1-1]

[0102] (Continuation of the table above) [Table 1-2]

[0103] As described above, in certain described embodiments, the call recalibration system 106 uses the above sequencing metrics to generate a set of machine learning predictions for different variants. In particular, in some embodiments, the call recalibration system 106 uses a call recalibration machine learning model to generate genotype probabilities (for SNPs) or uses another call recalibration machine learning model to generate variant call classifications (for indels) corresponding to various genomic coordinates. The call recalibration system 106 updates or corrects genotype calls by generating an updated genotype call data file, such as a variant call file (e.g., a recalibrated variant call file), based on the genotype probabilities and / or variant call classifications.

[0104] According to one or more embodiments, Figures 5A to 5C illustrate a call recalibration system 106 that generates either or both genotype probabilities and variant call classifications, generates genotype calls based on such likelihoods and / or classifications, and generates a recalibrated or updated variant call file containing genotype calls based on such probabilities and / or classifications. For example, Figure 5A illustrates a call recalibration system 106 according to one or more embodiments that uses a call recalibration machine learning model to generate genotype probabilities for (biallele) SNPs based on sequencing metrics corresponding to existing genotype calls. Figure 5B illustrates a call recalibration system 106 according to one or more embodiments that uses a call recalibration machine learning model to generate variant call classifications for indels (or variants other than biallele SNPs or biallele SNPs) based on sequencing metrics corresponding to existing genotype calls. Subsequently, Figure 5C illustrates a call recalibration system 106 that generates an updated variant call file containing recalibrated genotype calls based on genotype probability and / or variant call classification according to one or more embodiments.

[0105] As illustrated in Figure 5A, the call recalibration system 106 identifies the genome coordinate 502. For example, the call recalibration system 106 identifies the genome coordinate 502 from nucleobase calls corresponding to sample nucleotide sequences, or based on haplotype data corresponding to the genome coordinate 502. In some cases, the call recalibration system 106 identifies the genome coordinate 502 by (i) determining one or more nucleobase calls from nucleotide reads covering the genome coordinate, and (ii) determining that one or more nucleobase calls satisfy one or more threshold sequencing metrics (e.g., the base call quality metric of Q30). Additionally or alternatively, in certain embodiments, the call recalibration system 106 identifies the genome coordinate 502 from a database containing a haplotype reference panel correlated with a specific genome coordinate. Regardless of the identification method, in some cases, the call recalibration system 106 uses information from one or more sequencing data files 503, for example, information previously generated and stored by a call generation model (e.g., mutant cola as part of a call generation model), to identify the genomic coordinates 502.

[0106] In addition, the sequencing data file 503 accessed by the call recalibration system 106 also contains existing genotype calls (e.g., genotype calls generated by the call generation model and stored in the variant call file). More specifically, the sequencing data file 503 contains information determined by the call generation model (e.g., DRAGEN VC Caller) to generate existing genotype calls and predict the presence (or absence) of a variant (or specific genotype) at genomic coordinates 502. As described, the call generation model generates existing genotype calls by analyzing or processing sequencing metrics 504 (or subsets of sequencing metrics 504, such as read-based sequencing metrics and external source sequencing metrics), which are also available to the call recalibration system 106 from the sequencing data file 503. The call generation model also generates and stores (e.g., in the genotype call data file) some of the sequencing metrics 504 (e.g., call model-generated sequencing metrics) as part of predicting existing genotype calls.

[0107] In practice, the call recalibration system 106 extracts (or reconstructs) sequencing metrics 504 for genomic coordinates 502 from one or more existing sequencing data files. Specifically, the call recalibration system 106 extracts sequencing metrics associated with nucleotide reads generated by the call generation model or taken from another external source, as described above. Based on the sequencing metrics 504, the call recalibration system 106 further generates genotype probabilities 508 that together indicate a measure of confidence or probability that genomic coordinates 502 contain or exhibit a SNP. The genotype probabilities 508 represent an example of a variant call classification.

[0108] Specifically, as shown in Figure 5A, the call recalibration system 106 generates genotype probabilities 508 using the call recalibration machine learning model 506. For example, the call recalibration machine learning model 506 analyzes or processes the extracted sequencing metric 504 and the existing genotype call as input to generate genotype probabilities 508 as output, including: i) a first genotype probability 510 where the existing genotype call is a homozygous reference genotype (e.g., "L(0 / 0)@chr5:4") at genomic coordinates 502; ii) a second genotype probability 512 where the existing genotype call is a heterozygous mutant genotype (e.g., "L(0 / 1)@chr5:4") at genomic coordinates 502; and iii) a third genotype probability 514 where the existing genotype call is a homozygous mutant genotype (e.g., "L(1 / 1)@chr5:4") at genomic coordinates 502.

[0109] As mentioned, the call recalibration system 106 generates a genotype probability 508 to predict whether the SNP occurs at genomic coordinates 502. In some embodiments, to predict whether an indel occurs at genomic coordinates, the call recalibration system 106 generates a different set of machine learning predictions. Specifically, the call recalibration system 106 generates a variant call classification indicating the presence (or absence) of an indel (or another variant other than a multi-allele SNP or bi-allele SNP) at genomic coordinates in the sample sequence.

[0110] As shown in Figure 5B, the call recalibration system 106 generates a variant call classification 522 using the call recalibration machine learning model 520. More specifically, the call recalibration system 106 generates a variant call classification 522 using the call recalibration machine learning model 520 based on sequencing metrics 518 extracted (or reconstructed) from the sequencing data file 517 and existing genotype calls associated with genomic coordinates 516. For example, in some embodiments, the call recalibration system 106 identifies a biallelegenic genomic coordinate, such as genomic coordinate 516, and supplies the call recalibration machine learning model 520 with sequencing metrics for the biallelegenic genomic coordinate. In fact, similar to the above considerations regarding generating genotype probabilities for biallelegenic SNPs, the call recalibration system 106 similarly extracts (or reconstructs) sequencing metrics 518 associated with genomic coordinates 516, including read-based sequencing metrics, call model-generated sequencing metrics, and sequencing metrics from external sources. For example, the call recalibration system 106 analyzes a subset of sequencing metrics 518 (e.g., read-based sequencing metrics and / or external source sequencing metrics) extracted from a sequencing data file 517 to determine an existing genotype call (e.g., indicating a specific genotype or mutation at genomic coordinate 516). In some cases, when generating the sequencing data file 517, the call generation model generates a subset of sequencing metrics 518 (e.g., call model-generated sequencing metrics) associated with genomic coordinate 516.

[0111] In generating the mutation call classification 522 for genome coordinate 516, the call recalibration system 106 utilizes the call recalibration machine learning model 520. Specifically, the call recalibration system 106 utilizes the call recalibration machine learning model 520 to generate (i) the true positive mutation probability 524, which is that an existing genotype call (e.g., from the initial VCF file of the sequencing data file 517) is a true positive mutation call at genome coordinate 516; (ii) the zygosity error probability 528, which includes a zygosity error at genome coordinate 516; and (iii) the reference probability 532, which is that the existing genotype call at genome coordinate 516 is a homozygous reference genotype (or false positive). In some cases, the mutation call classifications 522 are mutually exclusive.

[0112] As further shown in Figure 5B, the true positive mutation probability 524 is represented by "TP". The symbol "TP" represents the probability that the input (x) is a true positive mutation in the existing genotype call data file of sequencing data file 517 (e.g., the initial VCF file of sequencing data file 517), where "TP" can be formulated as P(tp|x)). In contrast, the zygote error probability 528 is represented by "HH". The symbol "TP&HH" represents the probability that the input (x) is not a true positive but a het-hom error in the existing genotype call data file (e.g., the initial VCF file of sequencing data file 517). Furthermore, the reference probability 532 is represented by "FP", which indicates the probability that the input (x) is a false positive, and can be formulated as P(fp|x)).

[0113] To elaborate on the conjugation error probability 528, the call recalibration system 106 determines the probability that the predicted genotype at genomic coordinate 516 (e.g., an existing genotype call) is an inaccurate genotype (e.g., a genotype inaccurately identified by the call generation model) or contains an inaccurate allele. In some cases, the call recalibration system 106 determines, based on the sequencing metric 518, the probability that a conjugation error (e.g., a het / hom error) exists at genomic coordinate 516—for example, the substitution base is correct but the genotype is incorrect—or that the nucleobase call represents either the entire incorrect genotype or an incorrect allele in the existing genotype call. For example, when determining the probability that a zygote error exists, the call recalibration system 106 determines the probability that an alternative base call, represented as "1", is accurate but the genotype is inaccurate, such as the probability of inaccurately determining a 0 / 1 genotype call (e.g., A / T) instead of the accurate 1 / 1 genotype call (e.g., T / T) (or vice versa if the accurate genotype call is 0 / 1).

[0114] By determining the conjugation error probability 528, the call recalibration system 106 can correct the inaccuracies of existing sequencing systems where inaccurate calls are often indels. In particular, the call recalibration system 106 can more accurately generate genotype calls for genomic coordinates corresponding to indels when existing sequencing systems determine that genotype calls represent inaccurate genotypes representing inaccurate alleles resulting from long insertion or deletion sequences.

[0115] As further shown in Figure 5B, the call recalibration system 106 utilizes the call recalibration machine learning model 520 to generate the true positive mutation probability 524. In particular, the call recalibration system 106 generates the true positive mutation probability 524 based on the sequencing metric 518 for the existing genotype call at genomic coordinate 516. In some cases, the true positive mutation probability represents the probability of the correct mutant call genotype at genomic coordinate 516. For example, the call recalibration system 106 generates the probability that the existing genotype call for genomic coordinate 516, determined by the call generation model and stored in the sequencing data file 517, is correct.

[0116] Following Figure 5C, in some embodiments, the call recalibration system 106 uses the genotype probability 508 and / or the variant call classification 522 to update one or more data fields associated with the variant call file (e.g., the sequenced data file accessed by the call recalibration system 106), or the variant call file field ("VCF" field). For example, the call recalibration system 106 generates a recalibrated genotype call data file 536 (in this case, a recalibrated variant call file) based on the genotype probability 508. In some cases, the call recalibration system 106 generates a single merged variant call file that combines data from the genotype probability 508 for SNPs (e.g., updated genotype calls and / or updated sequenced metrics) and data from the variant call classification 522 for indels.

[0117] As shown, the call recalibration system 106 either displays updated sequencing metrics for existing genotype calls or generates corresponding updated VCF fields 534. Specifically, the call recalibration system 106 generates one set of updated VCF fields 534 based on genotype probabilities 508 for a set of genomic coordinates. Furthermore, the call recalibration system 106 generates another set of updated VCF fields 534 based on variant call classifications 522 for different sets of genomic coordinates. In some cases, the call recalibration system 106 modifies or updates only specific VCF fields based on genotype probabilities 508 and / or variant call classifications 522, leaving other fields unupdated.

[0118] In other cases, the call recalibration system 106 does not update the VCF field. When updating or correcting a genotype call, for example, the call recalibration system 106 does not update certain fields, such as the genotype (GT) field, based on the genotype probability 508 and / or variant call classification 522. In fact, in some cases, the call recalibration system 106 does not correct or update the GT field because there may not be enough information to determine the new or updated genotype in genomic coordinates.

[0119] To illustrate one embodiment, Figure 5C depicts a call recalibration system 106 that generates an updated VCF field 534 for 1 / 2 genotype (GT), where cytosine represents the reference base in genomic coordinates for the allele corresponding to the reference genome (indicated as "reference: C"), and thymine represents the first alternative base in genomic coordinates for the different allele ("alternative: T"). However, Figure 5C only shows examples of possible reference and alternative bases in genomic coordinates. The call recalibration system 106 generates genotype probabilities 508 and variant call classifications 522, and can modify the corresponding sequencing metrics of the VCF field for various other reference and alternative bases in genomic coordinates.

[0120] As further illustrated in Figure 5C, the call recalibration system 106 generates an updated base call quality metric in the base call quality (QUAL) field. More specifically, the call recalibration system 106 modifies or updates the base call quality metric based on genotype probabilities 508 and / or variant call classifications 522 to indicate the accuracy of the genotype call. As shown, the updated base call quality field shows a QUAL score of 48 for variants at the corresponding genomic coordinates. In this example, the updated base call quality metric (e.g., a QUAL score of 48) represents the score for any type of variant at the corresponding genomic coordinates. The call recalibration system 106 also generates a modified or updated genotype quality (GQ) field. For example, based on variant call classifications 522, the call recalibration system 106 generates a modified or updated genotype quality metric that indicates the likelihood or probability that the predicted genotype at the genomic coordinates is accurate. As shown, for example, the updated genotype quality field displays a genotype quality metric (e.g., a GQ score of 4 for 1 / 2 genotypes) for genotype calls that have heterozygous genotypes with respect to the multi-allele coordinate.

[0121] In one or more embodiments, the call recalibration system 106 further generates or updates a genotype probability field and (optionally) uses the genotype probability field to rank alleles. More specifically, the call recalibration system 106 generates an updated GT field by ordering candidate genotype calls to genomic coordinates according to their respective probabilities belonging to genomic coordinates 502. For example, the call recalibration system 106 determines the probability that each diploid genotype is associated with multiple genotypes composed of a pair of alleles. As another example, the call recalibration system 106 determines the relative probabilities associated with multiple alleles belonging to a multi-allele genomic coordinate (e.g., derived from a reference genome, a first alternative allele, and a second alternative allele).

[0122] In some embodiments, the Cole recalibration system 106 also, or alternatively, generates metrics for the PHRED-scaled likelihood (PL) field as part of the updated VCF field. For example, the Cole recalibration system 106 generates metrics for the PL field that can represent, for example, homozygous reference genotype, heterozygous genotype, and homozygous alternative genotype (each having PL field nomenclature 9 / 0 / 3).

[0123] In one or more embodiments, the call recalibration system 106 generates allele-specific probabilities or likelihoods based on the relative probabilities of genotype calls corresponding to alleles from the call generation model and other (non-referenced) genotypes identified by the call recalibration machine learning model. For example, in some embodiments, the call recalibration system 106 displays relative probability scores for each allele corresponding to each genotype call in a PL field showing normalized PHRED scale likelihood for the genotype, and / or a genotype probability (GL) field showing log-scale likelihood (e.g., log10 scale) for the called genotypes with given data (e.g., sequencing metrics).

[0124] As a motivation to modify a specific VCF field for a SNP, the call recalibration system 106 may, in some cases, utilize a call recalibration machine learning model to generate genotype probabilities 508 (whose sum to 1). In particular, the call recalibration machine learning model may generate a first genotype probability 510 of 0.1, a second genotype probability 512 of 0.2, and a third genotype probability 514 of 0.7. Based on the genotype probabilities 508 in such an example, the call recalibration system 106 generates updated genotype probability fields by updating the GT, GP, and PL fields using a combination of information from sequencing data files previously generated by the call recalibration machine learning model and the call generation model.

[0125] As further illustrated in Figure 5C, the Cole recalibration system 106 updates the PL fields for different genotypes (GTs). According to the normalization scale of PL scores, a relatively lower score for a genotype (e.g., PL0) indicates a relatively higher likelihood that the genotype exists in the genomic coordinates, while a relatively higher score for a genotype (e.g., PL101) indicates a relatively lower likelihood that the genotype exists in the genomic coordinates. For example, the Cole recalibration system 106 determines a PL score of 111 for the 0 / 0 genotype, a PL score of 52 for the 0 / 1 genotype, and a PL score of 52 for the 1 / 1 genotype. Thus, in Figure 5C, a PL score of 52 indicates the highest likelihood or selected genotype (e.g., the 0 / 1 and 1 / 1 genotypes), and a PL score of 111 indicates the lowest likelihood (e.g., the 0 / 0 genotype).

[0126] In some cases, the call recalibration system 106 generates an updated genotype probability field as a ranking of multiple alleles identified via the call generation model (without using the call recalibration machine learning model). In other cases, the call recalibration system 106 utilizes a special version of the call recalibration machine learning model, trained to generate an updated genotype probability field based on genotype probabilities 508 and / or variant call classifications 522.

[0127] As further shown in Figure 5C, the call recalibration system 106 generates or updates genotype call data files, such as the initial mutation call file, to create a recalibrated genotype call data file 536. For example, the call recalibration system 106 generates a recalibrated genotype call data file 536 from updated VCF fields 534 corresponding to the genotype probability 508 and mutation call classification 522, respectively. Thus, the call recalibration system 106 generates a recalibrated genotype call data file 536 for SNP genotype calls based on the genotype probability 508. As shown above, in some embodiments, the call recalibration system 106 generates a recalibrated genotype call data file that merges data for SNPs and indels from both the genotype probability 508 and the mutation call classification 522.

[0128] The call recalibration system 106 may generate a recalibrated genotype call data file 536 that includes an updated VCF field 534, which includes a base call quality metric, a genotype quality metric, and / or an updated genotype probability field. For example, the call recalibration system 106 selects a VCF field from an existing genotype call generated by a call generation model to include in the recalibrated genotype call data file. However, in some embodiments, the call recalibration system 106 does not select a field, but instead generates a new VCF field for the recalibrated genotype call data file by using a call recalibration machine learning model to process genotype probabilities 508 and variant call classifications 522.

[0129] As mentioned, in some cases, the call recalibration system 106 updates only specific fields, while other fields, such as the genotype (GT) field, remain unchanged. For example, the call recalibration system 106 updates the genotype quality field and the base call quality field. For other data fields, such as the PHRED scale likelihood (PL) and posterior genotype probability (GP) normalized for genotypes, the call recalibration system 106 either (i) keeps the field as is, (ii) removes the field, or (iii) updates the field to reflect the GQ for the called genotype and class 0 output 0 / 0. In some cases, the call recalibration system 106 maintains the relative probabilities of other genotypes with respect to the called genotype to ensure consistent updates and that the called genotype is the highest. In certain embodiments, the call recalibration system 106 maintains the distance of other genotypes from the called genotype by updating only the 0 / 0 and 1 / 2 values. By updating only specific fields, a call recalibration system can more efficiently generate (recalibrated and / or merged) genotype call data files without regenerating completely new genotype call data (as is done by some conventional systems) and / or without updating all fields (even those that are not changed by the new predictions).

[0130] Within (or as a result of) the recalibrated genotype call data file, the call recalibration system 106 may include or update one or more output genotype calls (e.g., variant calls) associated with genomic coordinates, as determined based on the updated VCF field 534. In fact, to generate updated genotype calls, the call recalibration system 106 may predict nucleo bases from candidate alleles at genomic coordinates (e.g., according to their respective probabilities and metrics, as indicated by the recalibrated variant call file). Thus, the call recalibration system 106 may generate a recalibrated genotype call data file 536 containing updated genotype calls for a particular genomic coordinate or confirmed genotype calls for a particular genomic coordinate. Similarly, the call recalibration system 106 may generate a recalibrated genotype call data file 536 containing one or more of the following for confirmed or modified genotype calls: modified base call quality metric, modified genotype probability metric, modified genotype metric, modified genotype likelihood metric, or modified genotype quality metric.

[0131] As mentioned above, in certain embodiments, the call recalibration system 106 trains or adjusts a call recalibration machine learning model (e.g., call recalibration machine learning models 418, 506, or 520). In particular, the call recalibration system 106 adapts the call recalibration machine learning model by utilizing an iterative training process to adjust or add decision trees or learning parameters that result in accurate variant call classifications (e.g., variant call classification 420, genotype probability 508, or variant call classification 522). For example, Figure 6 shows training a call recalibration machine learning model according to one or more embodiments.

[0132] As shown in Figure 6, the call recalibration system 106 accesses a sample sequencing data file 620 (e.g., a sequencing data file generated using an existing call generation model such as call generation model 204a), extracts (or reconstructs) sample sequencing metrics 604 from the sample sequencing data file 620, and receives or obtains several metrics (e.g., metrics from external sources) from a database 602 (e.g., database 116). For example, the call recalibration system 106 extracts (or reconstructs) sample sequencing metrics including sample read-based metrics, sample external source sequencing metrics, and sample call model-generated sequencing metrics.

[0133] As described above, in some embodiments, the call recalibration system 106 reconstructs at least some of the call model generation metrics that are not stored in the provided sample sequencing file but were utilized by the call generation model. For example, as shown in Figure 6, the call recalibration system 106 determines (i.e., derives) at least some of the call model generation metrics from alternative information in the sequencing data file in order to determine the reconstructed call model generation metrics. For example, in some implementations, the call recalibration system 106 reconstructs certain call model generation metrics, such as hidden Markov model (HMM) statistics, utilized by the call generation model, from other information in the sequencing data file, such as a concise and distinctive gapped alignment report (CIGAR) string output or other sequencing information.

[0134] In some cases, sample sequencing data files 620 have corresponding ground truth variant call files 616 associated with them, and the ground truth variant call files 616 show the actual genotype calls and their various metrics arising from the sample sequencing metrics 604. For example, the call recalibration system 106 utilizes ground truth variant call files from a training dataset from the Food and Drug Administration called the PrecisionFDA dataset. In some cases, the sample sequencing metrics 604 include a subset of sample sequencing metrics for each genotype call in the ground truth variant call file 616. The ground truth variant call file 616 may have ground truth variant calls (e.g., genotype metrics in the genotype field) and / or ground truth base calls corresponding to each subset of sample sequencing metrics.

[0135] As further illustrated in Figure 6, the call recalibration system 106 generates a predicted mutation call classification 608 based on the extracted sample sequencing metric 604. Specifically, the call recalibration system 106 generates the predicted mutation call classification 608 using a call recalibration machine learning model 606. In fact, in some embodiments, the call recalibration machine learning model 606 generates a set of three predicted mutation call classifications 608, including a predicted false positive probability, a predicted zygote error probability, and a predicted true positive classification. Thus, the predicted mutation call classification 608 can take any of the forms of mutation call classification described above.

[0136] Based on the predicted variant call classification 608, the call recalibration system 106 determines the genotype call and generates a modified variant call file 610 containing the modified or updated genotype call and corresponding fields. As described above, the call recalibration system 106 may modify the data fields corresponding to the variant call file (e.g., in the sample sequencing data file 620) for the genotype call using (i) existing genotype calls generated by the call generation model and contained in the sample sequencing data file 620, and (ii) the call recalibration machine learning model 606. Such modified or recalibrated values ​​are output to the modified variant call file 610, for example, by the call recalibration machine learning model 606. For example, the call recalibration system 106 determines recalibration values ​​for specific metrics in the modified variant call file 610, including the base call quality metric (QUAL), the genotype metric (GT), and the genotype-quality metric (GQ).

[0137] As further illustrated in Figure 6, the call recalibration system 106 performs a comparison 612. Specifically, the call recalibration system 106 performs a comparison 612 between (i) the variant genotype calls and / or data fields in the modified variant call file 610 and (ii) the variant genotype calls and / or data fields in the ground truth variant call file 616. In some embodiments, the call recalibration system 106 uses a loss function 614 to compare the variant genotype calls and / or data fields from the two variant call files (e.g., to determine a measure of error or loss between them). For example, if the call recalibration machine learning model 606 is an ensemble of gradient boosted trees, the call recalibration system 106 uses a mean squared error loss function (e.g., for regression) and / or a logarithmic loss function (e.g., for classification) as the loss function 614.

[0138] In contrast, in embodiments where the call recalibration machine learning model 606 is a neural network, the call recalibration system 106 may utilize a cross-entropy loss function, an L1 loss function, or a mean squared error loss function as the loss function 614. For example, the call recalibration system 106 uses the loss function 614 to determine the difference between the mutated genotype calls and / or data fields from the corrected mutated call file 610 and the ground truth mutated call file 616.

[0139] As further illustrated in Figure 6, the call recalibration system 106 performs model fitting 618. In particular, the call recalibration system 106 adapts the call recalibration machine learning model 606 based on comparison 612. For example, the call recalibration system 106 makes modifications or adjustments to the call recalibration machine learning model 606 to reduce the measure of loss from the loss function 614 for subsequent training iterations.

[0140] In the case of gradient-boosted trees or treelights, for example, the call recalibration system 106 trains a call recalibration machine learning model 606 on the gradient of errors determined by the loss function 614. For example, the call recalibration system 106 solves a convex optimization problem (e.g., infinite-dimensional) while regularizing the objective function to avoid overfitting. In certain implementation forms, the call recalibration system 106 scales the gradient to emphasize corrections for underrepresented classes (e.g., when there are significantly more true positive mutation calls than false positive mutation calls).

[0141] In some embodiments, the call recalibration system 106 adds a new weak learner (e.g., a new boosted tree) to the call recalibration machine learning model 606 with each successive training iteration as part of solving the optimization problem. For example, the call recalibration system 106 finds a feature (e.g., an array decision metric) that minimizes the loss from the loss function 614 and adds that feature to the tree for the current iteration, or starts building a new tree that has that feature.

[0142] In addition to, or as an alternative to, gradient-boosted decision trees, the Call recalibration system 106 trains logistic regressions to learn parameters for generating one or more variant Call classifications, such as true positive classifications. To avoid overfitting, the Call recalibration system 106 further regularizes based on hyperparameters such as learning rate, stochastic gradient boosting, number of trees, tree depth, complexity penaltyning, and L1 / L2 regularization.

[0143] In embodiments where the call recalibration machine learning model 606 is a neural network, the call recalibration system 106 performs model fitting 618 by modifying the internal parameters (e.g., weights) of the call recalibration machine learning model 606 to reduce the measure of loss for the loss function 614. In effect, the call recalibration system 106 modifies how the call recalibration machine learning model 606 analyzes and passes data between layers and neurons by modifying the internal network parameters. Thus, over multiple iterations, the call recalibration system 106 improves the accuracy of the call recalibration machine learning model 606.

[0144] In fact, in some cases, the call recalibration system 106 repeats the training process illustrated in Figure 6 multiple times. For example, the call recalibration system 106 repeats the training by selecting a new set of sequencing metrics for each genotype call, along with the corresponding ground truth genotype calls in the corresponding ground truth variant call file. The call recalibration system 106 further generates a new set of predicted variant call classifications with each iteration, along with the new modified variant call file. As described above, the call recalibration system 106 also performs model fitting 618 by comparing the variant genotype calls and / or data fields from the modified variant call file in each iteration with the corresponding variant genotype calls and / or data fields from the corresponding ground truth variant call file. The call recalibration system 106 repeats this process until the call recalibration machine learning model 606 generates a predicted variant call classification that results in variant calls that meet the loss threshold measure. In some embodiments, the call recalibration system 106 performs the training process shown in Figure 6 on homozygous reference coordinates to update or correct the variant calls for these coordinates, thereby recovering false negative variant calls (based on simulating haploid data from diploid data and correcting the inputs and outputs of the call recalibration machine learning model 606, as described).

[0145] As mentioned above, in certain described embodiments, the call recalibration system 106 provides improved computational efficiency compared to existing sequencing systems. For example, Figure 7 shows a table illustrating the improvements in computational efficiency associated with the call recalibration system in one or more embodiments.

[0146] As mentioned above, some existing sequencing systems that utilize machine learning model-based mutant cola must reprocess raw sequencing data (e.g., nucleotide reads from BCL files) to generate updated genotype calls for sample nucleotide sequences (e.g., genome samples). Furthermore, some existing sequencing systems require computer arrays with hardware accelerators such as field-programmable gate arrays (FPGAs) to run machine learning model-based mutant cola. In contrast, as discussed in detail above, the call recalibration system 106 recalibrates existing genotype calls based on sequencing metrics extracted or determined from one or more existing sequencing data files (e.g., sequencing data files generated by previous-generation call generation models) without reprocessing raw sequencing metrics (e.g., nucleotide reads from BCL files), utilizing a call recalibration machine learning model that can run on any processor.

[0147] As shown in the table in Figure 7, in at least one implementation, existing sequencing systems require an average of 20 minutes per genome sample to reprocess base call data for the corresponding genome sample. Such existing sequencing systems require 20 minutes per genome sample despite implementing a large FPGA-equipped computer with 48 parallel processing units and 256 GB of memory. As also shown in the table in Figure 7, in at least one implementation, the call recalibration system 106 utilizes a call recalibration machine learning model to process genome samples in an average of 7 minutes per genome sample to generate a recalibrated sequencing data file. By extracting sequencing metrics related to genotype calls from existing sequencing data files and utilizing the call recalibration machine learning model to generate variant call classifications that facilitate updating previously generated genotype calls and / or corresponding sequencing metrics, the call recalibration system 106 improves computer processing speed by more than double compared to existing sequencing systems performing the same basic tasks. Rather than limiting the call recalibration machine learning model to FPGAs or other reconfigurable arrays, the call recalibration system 106 extends its implementation to any processor, such as a general-purpose CPU array with 16 processors and 128GB of memory, as shown in Figure 7. In fact, as shown by the comparative experimental results provided in Figure 7, the call recalibration system 106 provides a significant improvement in efficiency over existing sequencing systems.

[0148] As mentioned above, in certain described embodiments, the call recalibration system 106 offers improved efficiency and accuracy compared to existing sequencing systems. Figures 8A and 8B show bar graphs illustrating the accuracy improvements associated with the call recalibration system 106 in one or more embodiments. In particular, Figures 8A and 8B show comparative experimental results of various sequencing systems performing various sample genome sequences.

[0149] For example, Figure 8A shows a bar graph comparing performance in identifying single nucleotide variants (SNPs) in various genome datasets (i.e., portions of the HG001–HG007 human genome datasets). The bar graph shows the results for an earlier version of the call generation model labeled "v3.7.8" (e.g., call generation model 204a), an updated version of the call generation machine learning model labeled "v4.0.3" (e.g., updated call generation model 204b), and a call recalibration system 106 labeled "v4.2" that utilizes a call recalibration machine learning model to recalibrate nucleotide-based reads generated by the earlier version of the call generation model.

[0150] As shown in Figures 8A and 8B, the call recalibration system 106 performs better than the previous call generation model, resulting in fewer false positives (FP) and false negatives (FN) when identifying SNPs within each genome dataset. Furthermore, the call recalibration system 106 performs similarly to, and more efficiently than, the updated call generation model using the call recalibration machine learning model (as discussed above, for example, with respect to Figure 7). The call recalibration system 106 also provides similar results when using alternative formats for sequencing data files. As shown, the FP and FN results for SNP calls are similar when using BAM files and when using CRAM files.

[0151] Similar to Figure 8A, Figure 8B shows a bar graph comparing performance in identifying mutations, including insertions or deletions (indels), within various genome datasets (i.e., portions of the HG001–HG007 human genome datasets). The bar graphs shown represent the results for an earlier version of the call generation model labeled "v3.7.8" (e.g., call generation model 204a), an updated version of the call generation machine learning model labeled "v4.0.3" (e.g., updated call generation model 204b), and a call recalibration system 106 labeled "v4.2," which utilizes a call recalibration machine learning model to recalibrate nucleotide-based reads generated by the earlier version of the call generation model.

[0152] As shown, the call recalibration system 106 performs better than the previous call generation model, resulting in fewer false positives (FP) and false negatives (FN) when identifying indels within each genome dataset. Furthermore, the call recalibration system 106 performs similarly to, and more efficiently than, the updated call generation model using the call recalibration machine learning model (as discussed above with respect to Figure 7, for example). The call recalibration system 106 also provides similar results when using alternative formats for sequencing data files. As shown, the FP and FN results for indel calls are similar when using BAM files as input to determine the sequencing metric and when using CRAM files as input to determine the sequencing metric.

[0153] Referring now to Figure 9, this figure shows an exemplary flowchart of a set of operations that generate a recalibrated sequencing data file having updated genotype calls or mutation calls based on mutation call classification from a call recalibration machine learning model according to one or more embodiments. Figure 9 shows operations according to one embodiment, but alternative embodiments may omit, add, rearrange, and / or modify any of the operations shown in Figure 9. The operations in Figure 9 may be performed as part of a method. Alternatively, the non-temporary computer-readable storage medium may contain instructions, which, when executed by one or more processors, cause a computing device to perform the operations shown in Figure 9. In further embodiments, the system comprises at least one processor and a non-temporary computer-readable medium containing instructions, which, when executed by one or more processors, cause the system to perform the operations in Figure 9.

[0154] As shown in Figure 9, the sequence of operations 900 includes operations 902 to access the sequencing data file, 904 to extract sequencing metrics for genotype calls, 906 to generate variant call classifications for genotype calls, and 908 to generate a recalibrated sequencing data file. For example, the sequence of operations 900 may include operations that perform any of the tasks described in the following clauses. Clause 1. Method, Access to one or more sequencing data files containing data on nucleotide reads and genotype calls in genomic coordinates for the sample nucleotide sequence, Extracting sequencing metrics for nucleotide reads or genotype calls from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics, generate one or more variant call classifications that indicate the accuracy of genotype calls in one or more sequencing data files, A method comprising generating a recalibrated sequencing data file containing updated genotype calls in genomic coordinates for a sample nucleotide sequence based on one or more mutation call classifications. Clause 2. The method according to Clause 1, wherein generating one or more mutant call classifications includes generating one or more mutant call classifications without using a call generation model to simultaneously generate genotype calls. Clause 3. The method according to Clause 1 or 2, wherein extracting sequencing metrics for genotype calls includes extracting one or more read-based sequencing metrics from one or more sequencing data files or alignment data files. Clause 4. The method according to any one of Clauses 1 to 3, wherein extracting sequencing metrics includes extracting one or more read-based sequencing metrics or call model generation sequencing metrics for genotype calls from a genotype call data file of one or more sequencing data files, the genotype call data file containing genotype calls. Clause 5. The method described in any one of Clauses 1 to 4, further comprising accessing a genotype call data file by accessing a mutation call format (VCF) file or a genomic mutation call format (gVCF) file containing mutation calls and non-mutation calls. Clause 6. The method according to any one of Clauses 1 to 5, wherein the genotype call data file containing the genotype call is generated on a computing device running a hardware accelerator, and the recalibrated sequencing data file is generated using a general-purpose processing unit as at least one processor in the system. Clause 7. The method described in any one of Clauses 1 to 6, wherein the hardware accelerator includes a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), and the recalibrated sequence determination data file is generated using one or more of a central processing unit (CPU) or a graphical processing unit (GPU). Article 8. Accessing additional genotype call data files generated by a different version of the call generation model than the version that generated the genotype call data file, Extracting additional sequencing metrics for additional genotype calls in genomic coordinates for additional sample nucleotide sequences from additional genotype call data files, Using a call recalibration machine learning model and based on additional sequencing metrics for additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of additional genotype calls in an additional genotype call data file. The method according to any one of Clauses 1 to 7, further comprising generating an additional recalibrated sequencing data file containing updated additional genotype calls in genomic coordinates for additional sample nucleotide sequences, based on one or more additional mutation call classifications. Clause 9. The method of any one of Clauses 1 to 8, further comprising generating one or more variant call classifications by generating one or more of the following: a false positive probability that the genotype call is false positive, a genotype error probability that the genotype of the genotype call is incorrect, or a true positive probability that the genotype call is true positive. Article 10. The genome coordinates were identified as the multi-allele genome coordinates. Using a call recalibration machine learning model, one or more variant call classifications are generated in which the genotype call includes one or more of the following: a reference probability in which the genotype call contains a homozygous reference genotype in a multi-allele genome coordinate system, a zygosity error probability in which the genotype call contains a zygosity error in a multi-allele genome coordinate system, or a true positive mutation probability in which the genotype call constitutes a true positive mutation in a multi-allele genome coordinate system. The method according to any one of Clauses 1 to 9, further comprising determining the updated genotype call by determining the updated genotype call in the multi-allele genomic coordinates based on one or more of the reference probability, the zygote error probability, or the true positive mutation probability. Article 11. Modify one or more of the base call quality metrics, genotype probability metrics, genotype metrics, genotype likelihood metrics, or genotype quality metrics for genotype calls based on one or more variant call classifications. The method according to any one of the clauses 1 to 10, further comprising generating a recalibrated sequencing data file including a modified base call quality metric, a modified genotype probability metric, a modified genotype metric, a modified genotype likelihood metric, or a modified genotype quality metric. Article 12. Determining homozygous reference genotype calls in genomic coordinates instead of heterozygous or homozygous mutant genotype calls reported in one or more sequencing data files. Instead of determining the heterozygous mutant genotype call in genomic coordinates, instead of the homozygous reference genotype call or homozygous mutant genotype call reported in one or more sequencing data files, The method according to any one of the claims 1 to 11, further comprising generating updated genotype calls in biallelic genomic coordinates for a sample nucleotide sequence as part of a recalibrated sequencing data file by determining homozygous mutant genotype calls in genomic coordinates instead of heterozygous mutant genotype calls or homozygous reference genotype calls reported in one or more sequencing data files. Article 13. Extracting sequencing metrics for additional genotype calls at additional genomic coordinates for a sample nucleotide sequence from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics for additional genotype calls, generate one or more additional variant call classifications that indicate the accuracy of additional genotype calls in one or more sequencing data files, Based on one or more additional variant call classifications, the base call quality metric for additional genotype calls is modified to generate a modified base call quality metric that falls below the base call quality threshold. The method described in any one of Clauses 1 to 12, further including annotating additional genotype calls to indicate that the modified base call quality metric falls below the base call quality threshold. Article 14. Extracting sequencing metrics for additional genotype calls at additional genomic coordinates for a sample nucleotide sequence from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics for additional genotype calls, generate one or more additional variant call classifications that indicate the accuracy of additional genotype calls in one or more sequencing data files, The method according to any one of the clauses 1 to 13, further comprising confirming genotype calls in additional genomic coordinates for a sample nucleotide sequence based on one or more additional mutation call classifications. Article 15. Generate multiple recalibrated sequencing data files from multiple sequencing data files corresponding to multiple known genomes. Updated genotype calls from multiple recalibrated sequencing data files are compared with known variations in multiple known genomes. The method described in any one of Clauses 1 to 14, further comprising training a call recalibration machine learning model by adjusting the parameters of the call recalibration machine learning model based on the difference between updated genotype calls and known mutations. Article 16. Method, Access to one or more sequencing data files containing genotype calls in genomic coordinates for the sample nucleotide sequence, Extracting sequencing metrics for genotype calls from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics, generate one or more variant call classifications that indicate the accuracy of genotype calls in one or more sequencing data files, A method comprising generating a recalibrated sequencing data file containing updated genotype calls in genomic coordinates for a sample nucleotide sequence based on one or more mutation call classifications. Article 17. Accessing alignment data files of one or more sequencing data files, wherein the alignment data files contain nucleotide reads corresponding to genomic coordinates for sample nucleotide sequences. The method according to Clause 16, further comprising extracting one or more read-based sequencing metrics from an alignment data file, wherein the one or more read-based sequencing metrics correspond to nucleotide reads. Clause 18. The method according to Clause 16 or 17, further comprising generating one or more variant call classifications without utilizing a call generation model to simultaneously generate genotype calls. Clause 19. The method according to any one of Clauses 16 to 18, wherein extracting sequencing metrics includes extracting one or more read-based sequencing metrics or call model generation sequencing metrics for genotype calls from a genotype call data file of one or more sequencing data files, the genotype call data file containing genotype calls. Clause 20. The method described in any one of Clauses 16 to 19, further comprising accessing a genotype call data file by accessing a mutation call format (VCF) file or a genomic mutation call format (gVCF) file containing mutation calls and non-mutation calls. Clause 21. The method according to any one of Clauses 16 to 20, wherein the genotype call data file containing genotype calls is generated on a computing device running a hardware accelerator, and the recalibrated sequencing data file is generated using a general-purpose processing unit as at least one processor in the system. Clause 22. The method described in any one of Clauses 16 to 21, wherein the hardware accelerator includes a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), and the recalibrated sequence determination data file is generated using one or more of a central processing unit (CPU) or a graphical processing unit (GPU). Article 23. Accessing additional genotype call data files generated by a different version of the call generation model than the version that generated the genotype call data file, Extracting additional sequencing metrics for additional genotype calls in genomic coordinates for additional sample nucleotide sequences from additional genotype call data files, Using a call recalibration machine learning model and based on additional sequencing metrics for additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of additional genotype calls in an additional genotype call data file. The method according to any one of the claims 16 to 22, further comprising generating an additional recalibrated sequencing data file containing updated additional genotype calls in genomic coordinates for additional sample nucleotide sequences, based on one or more additional mutation call classifications. Clause 24. The method according to any one of Clauses 16 to 23, further comprising, when executed by at least one processor, generating one or more variant call classifications for the system by generating one or more of the following: a false positive probability that a genotype call is false positive, a genotype error probability that the genotype of a genotype call is incorrect, or a true positive probability that a genotype call is true positive. Article 25. The genome coordinates were identified as the multi-allele genome coordinates. Using a call recalibration machine learning model, one or more variant call classifications are generated in which the genotype call includes one or more of the following: a reference probability in which the genotype call contains a homozygous reference genotype in a multi-allele genome coordinate system, a zygosity error probability in which the genotype call contains a zygosity error in a multi-allele genome coordinate system, or a true positive mutation probability in which the genotype call constitutes a true positive mutation in a multi-allele genome coordinate system. The method according to any one of the claims 16 to 24, further comprising determining the updated genotype call by determining the updated genotype call in the multi-allele genomic coordinates based on one or more of the reference probability, the zygote error probability, or the true positive mutation probability. Article 26. Modify one or more of the base call quality metrics, genotype probability metrics, genotype metrics, genotype likelihood metrics, or genotype quality metrics for genotype calls based on one or more variant call classifications. The method according to any one of the clauses 16 to 25, further comprising generating a recalibrated sequencing data file including a modified base call quality metric, a modified genotype probability metric, a modified genotype metric, a modified genotype likelihood metric, or a modified genotype quality metric. Article 27. Determining homozygous reference genotype calls in genomic coordinates instead of heterozygous or homozygous mutant genotype calls reported in one or more sequencing data files. Instead of determining the heterozygous mutant genotype call in genomic coordinates, instead of the homozygous reference genotype call or homozygous mutant genotype call reported in one or more sequencing data files, The method according to any one of the claims 16 to 26, further comprising generating updated genotype calls in biallelic genomic coordinates for a sample nucleotide sequence as part of a recalibrated sequencing data file by determining homozygous mutant genotype calls in genomic coordinates instead of heterozygous mutant genotype calls or homozygous reference genotype calls reported in one or more sequencing data files. Article 28. Extracting sequencing metrics for additional genotype calls at additional genomic coordinates for a sample nucleotide sequence from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics for additional genotype calls, generate one or more additional variant call classifications that indicate the accuracy of additional genotype calls in one or more sequencing data files, Based on one or more additional variant call classifications, the base call quality metric for additional genotype calls is modified to generate a modified base call quality metric that falls below the base call quality threshold. The method described in any one of the clauses 16 to 27, further including annotating additional genotype calls to indicate that the modified base call quality metric falls below the base call quality threshold. Article 29. Extracting sequencing metrics for additional genotype calls at additional genomic coordinates for a sample nucleotide sequence from one or more sequencing data files, Using a call recalibration machine learning model and based on sequencing metrics for additional genotype calls, generate one or more additional variant call classifications that indicate the accuracy of additional genotype calls in one or more sequencing data files, The method according to any one of the claims 16 to 28, further comprising confirming genotype calls in additional genomic coordinates for a sample nucleotide sequence based on one or more additional mutation call classifications. Article 30. Generate multiple recalibrated sequencing data files from multiple sequencing data files corresponding to multiple known genomes. Updated genotype calls from multiple recalibrated sequencing data files are compared with known variations in multiple known genomes. The method according to any one of the clauses 16 to 29, further comprising training a call recalibration machine learning model by adjusting the parameters of the call recalibration machine learning model based on the difference between updated genotype calls and known mutations. Article 31. Generating one or more variant call classifications involves generating variant call classifications for one or more candidate insertions or deletions (indels) using a call recalibration machine learning model trained on indel training data. The method according to any one of the claims 16 to 30, wherein generating a recalibrated sequencing data file includes generating an updated genotype call indicating the presence or absence of an indel in genomic coordinates for a sample nucleotide sequence, based on one or more mutation call classifications for one or more candidate indels. Article 32. Generating one or more mutation call classifications involves generating mutation call classifications for one or more candidate SNVs using a call recalibration machine learning model trained on single nucleotide variant (SNV) training data. The method according to any one of the claims 16 to 31, wherein generating a recalibrated sequencing data file includes generating an updated genotype call indicating the presence or absence of an SNV in genomic coordinates for a sample nucleotide sequence, based on one or more mutation call classifications for one or more candidate SNVs.

[0155] The methods described herein can be used in conjunction with various nucleic acid sequencing techniques. Particularly applicable techniques involve attaching nucleic acids to fixed positions within an array so that their relative positions do not change, and the array being repeatedly imaged. For example, embodiments in which images are obtained in different color channels corresponding to different labels used to distinguish one nucleotide base type from another are particularly applicable. In some embodiments, the process of determining the nucleotide sequence of a target nucleic acid may be an automated process. Preferred embodiments include synthetic sequencing (SBS) techniques.

[0156] SBS technology generally involves the enzymatic elongation of a nascent nucleic acid chain by the repeated addition of nucleotides to a template chain. In conventional SBS methods, a single nucleotide monomer can be delivered to the target nucleotide in the presence of polymerase at each delivery. However, the method described herein allows for the delivery of two or more types of nucleotide monomers to the target nucleic acid in the presence of polymerase during delivery.

[0157] SBS can utilize nucleotide monomers with a terminator moiety, or nucleotide monomers lacking any terminator moiety. Methods utilizing nucleotide monomers lacking a terminator include, for example, pyrosequencing and sequencing using γ-phosphate-labeled nucleotides, as described in further detail below. In methods using nucleotide monomers without a terminator, the number of nucleotides added in each cycle is generally variable and depends on the template sequence and the mode of nucleotide delivery. In SBS techniques utilizing nucleotide monomers with a terminator moiety, the terminator may be effectively irreversible under the sequencing conditions used, as in conventional Sanger sequencing using dideoxynucleotides, or it may be reversible, as in sequencing methods developed by Solexa (now Illumina, Inc.).

[0158] SBS technology can use nucleotide monomers with or without a labeled moiety. Therefore, integration events can be detected based on the characteristics of the label, such as fluorescence; the characteristics of the nucleotide monomer, such as molecular weight or charge; and by-products of nucleotide integration, such as pyrophosphate release. In embodiments where two or more different nucleotides are present in the sequencing reagent, the different nucleotides may be distinguishable from each other, or alternatively, two or more different labels may be distinguishable under the detection technique used. For example, different nucleotides present in the sequencing reagent may have different labels, and they can be distinguished using a suitable optical system, as exemplified by the sequencing method developed by Solexa (now Illumina, Inc.).

[0159] A preferred embodiment is pyrosequencing technology. Pyrosequencing detects the release of inorganic pyrophosphate (PPi) when specific nucleotides are incorporated into the nascent DNA chain (Ronaghi, M., Karamohamed, S., Pettersson, B., Uhlen, M. and Nyren, P. (1996) "Real-time DNA sequencing using detection of pyrophosphate release." Analytical Biochemistry 242(1), 84-9; Ronaghi, M. (2001) "Pyrosequencing sheds light on DNA sequencing." Genome Res. 11(1), 3-11; Ronaghi, M., Uhlen, M. and Nyren, P. (1998) "A sequencing method based on real-time pyrophosphate." Science 281(5375),363, U.S. Patent No. 6,210,891, U.S. Patent No. 6,258,568, and U.S. Patent No. 6,274,320 (the entire disclosure thereof is incorporated herein by reference). In pyrosequencing, emitted PPi can be detected by immediate conversion to adenosine triphosphate (ATP) by ATP sulfurase, and the level of the generated ATP is detected via photons produced by luciferase. The nucleic acids to be sequenced can be attached to features in an array, and the array can be imaged to capture the chemiluminescent signal generated by incorporating the nucleotides into the array features. After processing the array with a specific nucleotide type (e.g., A, T, C, or G), an image can be obtained. The images obtained after the addition of each nucleotide type differ in terms of which features in the array are detected. These differences in the image reflect the different sequence content of the features on the array. However, the relative position of each feature remains unchanged in the image. The images can be stored, processed, and analyzed using the methods described herein.For example, images obtained after processing an array with each different nucleotide type can be processed in the same manner as images obtained from different detection channels for a reversible terminator-based sequencing method, as illustrated herein.

[0160] In another exemplary type of SBS, cyclic sequencing is achieved by stepwise addition of reversible terminator nucleotides containing cleavable or photobleachable dye labels, such as those described in International Publication No. 04 / 018497 and U.S. Patent No. 7,057,026, whose disclosure is incorporated herein by reference. This technique has been commercialized by Solexa (now Illumina Inc.) and is also described in International Publication No. 91 / 06678 and International Publication No. 07 / 123,744, each of which is incorporated herein by reference. The availability of fluorescently labeled terminators with both ends reversible and the fluorescent label cleaved facilitates efficient cyclic reversible termination (CRT) sequencing. Polymerases can also be co-operated to efficiently incorporate and extend these modified nucleotides.

[0161] Preferably, in reversible terminator-based sequencing embodiments, the label does not substantially inhibit extension under SBS reaction conditions. However, the detection label may be removable, for example, by cleavage or degradation. Images can be captured after the incorporation of the label into the arrayed nucleic acid features. In certain embodiments, each cycle involves the simultaneous delivery of four different nucleotide types to the array, each nucleotide type having a spectrally different label. Four images can then be obtained by using a selective detection channel for each of the four different labels. Alternatively, different nucleotide types can be added sequentially, and an image of the array can be obtained between each addition step. In such embodiments, each image shows a nucleic acid feature incorporating a particular type of nucleotide. Because the sequence content of each feature part is different, different feature parts may or may not be present in different images. However, the relative positions of the features remain unchanged within the images. Images obtained from such a reversible terminator SBS method can be stored, processed, and analyzed as described herein. Following the image capture step, the label can be removed, and the reversible terminator portion can be removed for subsequent nucleotide addition and detection cycles. Removing the label after detection in a specific cycle and before subsequent cycles has the advantage of reducing background signal and crosstalk between cycles. Examples of useful labeling and removal methods are listed below.

[0162] In certain embodiments, some or all of the nucleotide monomers may include a reversible terminator. In such embodiments, the reversible terminator / cleavable fluorophore (fluor) may include a fluorophore (fluor) attached to the ribose moiety via a 3' ester bond (Metzker, Genome Res. 15:1767-1776 (2005), which is incorporated herein by reference). Other methods separate the chemistry of the terminator from the cleavage of the fluorescent label (this is incorporated herein by reference in its entirety, Ruparel et al., Proc Natl Acad Sci USA 102:5932-7 (2005)). Ruparel et al. describe the development of a reversible terminator that uses a small amount of 3' allyl group to block elongation but can be readily deblocked by short-term treatment with a palladium catalyst. The fluorophore was attached to the base via a photocleavable linker that can be readily cleaved by 30 seconds of exposure to long-wavelength UV light. Therefore, either disulfide reduction or photocleavage can be used as a cleavable linker. Another method to reversible termination is the use of a natural termination followed by the placement of a bulky dye on the dNTP. The presence of a charged bulky dye on the dNTP can act as an effective terminator via steric and / or electrostatic hindrance. The presence of one incorporation event prevents further binding unless the dye is removed. Cleavage of the dye removes the fluorophore (fluor) and effectively reverses the termination. Examples of modified nucleotides are also described in U.S. Patents 7,427,673 and 7,057,026, and these disclosures are incorporated herein by reference in their entirety.

[0163] Additional exemplary SBS systems and methods that can be used in conjunction with the methods and systems described herein are described in U.S. Patent Publication No. 2007 / 0166705, U.S. Patent Publication No. 2006 / 0188901, U.S. Patent No. 7,057,026, U.S. Patent Publication No. 2006 / 0240439, U.S. Patent Publication No. 2006 / 0281109, International Publication No. 05 / 065814, U.S. Patent Publication No. 2005 / 0100900, International Publication No. 06 / 064199, International Publication No. 07 / 010,251, U.S. Patent Publication No. 2012 / 0270305, and U.S. Patent Publication No. 2013 / 0260372, the disclosures of which are incorporated herein by reference in their entirety.

[0164] Several embodiments can utilize the detection of four different nucleotides using fewer than four different labels. For example, SBS can be carried out using the method and system described in the incorporated material, U.S. Patent Application Publication No. 2013 / 0079232. As a first example, pairs of nucleotide types can be detected at the same wavelength but can be distinguished based on differences in intensity for one member of the pair, or based on changes to one member of the pair (e.g., through chemical modification, photochemical modification, or physical modification) that cause a noticeable signal to appear or disappear compared to the signals detected for the other members of the pair. As a second example, three of the four different nucleotide types can be detected under specific conditions, while a fourth nucleotide type has no detectable label under those conditions or is minimally detectable under those conditions (e.g., minimal detection by background fluorescence). Incorporation of the first three nucleotide types into the nucleic acid can be determined based on the presence of their respective signals, and incorporation of the fourth nucleotide type into the nucleic acid can be determined based on the absence or minimal detection of any signal. As a third example, one nucleotide type may include a label that is detected by two different channels, while other nucleotide types are detected by one or fewer channels. The three exemplary configurations described above are not considered mutually exclusive and can be used in various combinations.An exemplary embodiment combining all three examples is a fluorescence-based SBS method using a first nucleotide type detected in a first channel (e.g., dATP with a label detectable in the first channel when excited by a first excitation wavelength), a second nucleotide type detected in a second channel (e.g., dCTP with a label detectable in the second channel when excited by a second excitation wavelength), a third nucleotide type detected in both the first and second channels (e.g., dTTP with at least one label detectable in both channels when excited by the first and / or second excitation wavelengths), and a fourth nucleotide type that is not detected in any channel or has minimally detected labels (e.g., unlabeled dGTP).

[0165] Furthermore, as described in the incorporated material, U.S. Patent Application Publication No. 2013 / 0079232, sequencing data can be obtained using a single channel. In such a so-called one-dye sequencing method, a first nucleotide type is labeled but the label is removed after the first image is generated, and a second nucleotide type is labeled only after the first image is generated. A third nucleotide type retains its label in both the first and second images, and a fourth nucleotide type remains unlabeled in both images.

[0166] Several embodiments can utilize sequencing by ligation techniques. Such techniques utilize DNA ligases to incorporate oligonucleotides and identify the incorporation of such oligonucleotides. Oligonucleotides typically have different labels that correlate with the identity of specific nucleotides in the sequence into which the oligonucleotide hybridizes. As with other SBS methods, an image can be obtained after processing an array of nucleic acid features with a labeled sequencing reagent. Each image shows a nucleic acid feature with a specific type of label incorporated. Because the sequence content of each feature region is different, different images may or may not have different feature regions, but the relative positions of the feature regions remain unchanged within the image. Images obtained from ligation-based sequencing methods can be stored, processed, and analyzed as described herein. Exemplary SBS systems and methods that can be used in conjunction with the methods and systems described herein are described in U.S. Patents 6,969,488, 6,172,218, and 6,306,597, and these disclosures are incorporated herein by reference in their entirety.

[0167] Several embodiments can utilize nanopore sequencing (Deamer, DW & Akeson, M. "Nanopores and nucleic acids: prospects for ultrarapid sequencing." Trends Biotechnol. 18, 147-151 (2000); Deamer, D. and D. Branton, "Characterization of nucleic acids by nanopore analysis." Acc. Chem. Res. 35: 817-825 (2002); Li, J., M. Gershow, D. Stein, E. Brandin, and J. Golovchenko, "DNA molecules and configurations in a solid-state nanopore microscope." Nat. Mater. 2: 611-615 (2003). These disclosures are incorporated herein by reference in their entirety). In such embodiments, the target nucleic acid passes through the nanopore. The nanopore may be a synthetic pore such as α-hemolysin or a biological membrane protein. As target nucleic acids pass through nanopores, each base pair can be identified by measuring the variation in the electrical conductance of the pores. (U.S. Patent No. 7,001,792; Soni, GV & Meller, "A. Progress toward ultrafast DNA sequencing using solid-state nanopores." Clin. Chem. 53, 1996-2001 (2007); Healy, K., "Nanopore-based single-molecule DNA analysis." Nanomed. 2, 459-481 (2007); Cockroft, SL, Chu, J., Amorin, M. & Ghadiri, MR, "A single-molecule nanopore device detects DNA polymerase activity with single-nucleotide resolution." J. Am Chem. Soc. 130, 818-820 (2008). These disclosures are incorporated herein by reference in their entirety.)Data obtained from nanopore arrangement determination can be stored, processed, and analyzed as described herein. Specifically, the data can be processed as images according to exemplary processing of optical and other images as described herein.

[0168] Some embodiments may utilize methods involving real-time monitoring of DNA polymerase activity. Nucleotide incorporation can be detected, for example, via fluorescence resonance energy transfer (FRET) interactions between fluorophore-containing polymerases and γ-phosphate-labeled nucleotides, as described in U.S. Patents 7,329,492 and 7,211,414, each incorporated herein by reference; or nucleotide incorporation can be detected using zero-mode waveguides, as described in U.S. Patent 7,315,019, each incorporated herein by reference, and fluorescent nucleotide analogs and manipulated polymerases, as described in U.S. Patent 7,405,281 and U.S. Patent Application Publication 2008 / 0108082, each incorporated herein by reference. Illumination can be limited to a zeptolite-scale volume around the surface-tethered polymerase so that the incorporation of fluorescently labeled nucleotides can be observed with low background (Levene, M.J. et al. "Zero-mode waveguides for single-molecule analysis at high concentrations." Science, 299, 682-686 (2003); Lundquist, P. et al. "Parallel confocal detection of single molecules in real time." Opt. Lett. 33, 1026-1028 (2008); Korlach, J. et al. "Selective aluminum passivation for targeted immobilization of single DNA polymerase molecules in zero-mode waveguide nanostructures." Proc. Natl. Acad. Sci. USA 105, 1176-1181 (2008). These disclosures are incorporated herein by reference in their entirety). Images obtained from such methods can be stored, processed, and analyzed as described herein.

[0169] Some SBS embodiments include the detection of protons released during the incorporation of nucleotides into the extension product. For example, sequencing based on the detection of released protons may use electrodetectors and related technologies commercially available from Ion Torrent (Guilford, CT, a subsidiary of Life Technologies), or sequencing methods and systems described in U.S. Patent Publication No. 2009 / 0026082(A1), U.S. Patent Publication No. 2009 / 0127589(A1), U.S. Patent Publication No. 2010 / 0137143(A1), or U.S. Patent Publication No. 2010 / 0282617(A1), each of which is incorporated herein by reference. The methods herein for amplifying target nucleic acids using dynamic exclusion can be readily applied to substrates used for proton detection. More specifically, the methods herein can be used to generate a clonal population of amplicons used for proton detection.

[0170] The SBS method described above can be advantageously implemented in a multiplex format so that multiple different target nucleic acids are manipulated simultaneously. In certain embodiments, different target nucleic acids can be processed on the surface of a common reaction vessel or a specific substrate. This allows for convenient delivery of sequencing reagents, removal of unreacted reagents, and detection of incorporation events in a multiplex format. In embodiments using surface-bound target nucleic acids, the target nucleic acids may be in array form. In array form, the target nucleic acids can typically be bound to the surface in a spatially distinguishable manner. The target nucleic acids can be bound by direct covalent bonding, binding to beads or other particles, or binding to polymerase or other molecules bound to the surface. The array may contain a single copy of the target nucleic acid at each site (also referred to as a feature), or multiple copies having the same sequence may be present at each site or feature. Multiple copies can be generated by amplification methods such as bridge amplification or emulsion PCR, which are described in more detail below.

[0171] The method described herein can use arrays having any of the following densities of feature parts: for example, at least about 10 feature parts / cm², 100 feature parts / cm², 500 feature parts / cm², 1,000 feature parts / cm², 5,000 feature parts / cm², 10,000 feature parts / cm², 50,000 feature parts / cm², 100,000 feature parts / cm², 1,000,000 feature parts / cm², 5,000,000 feature parts / cm², or more.

[0172] An advantage of the methods described herein is that they provide the rapid and efficient parallel detection of multiple target nucleic acids. Therefore, this disclosure provides an integrated system that allows for the preparation and detection of nucleic acids using techniques known in the art, such as those exemplified above. Accordingly, the integrated system of this disclosure may include a fluid component capable of delivering amplification reagents and / or sequencing reagents to one or more immobilized DNA fragments, and the system may include components such as pumps, valves, reservoirs, and fluid lines. A flow cell may constitute and / or be used in the integrated system for detecting target nucleic acids. Exemplary flow cells are described, for example, in U.S. Patent No. 2010 / 0111768(A1) and U.S. Patent Application No. 13 / 273666, each of which is incorporated herein by reference. As exemplified with respect to flow cells, one or more fluid components of the integrated system may be used in amplification and detection methods. Taking an embodiment of nucleic acid sequencing as an example, one or more fluid components of the integrated system may be used for the delivery of sequencing reagents in the amplification method described herein and in sequencing methods such as those exemplified above. Alternatively, the integrated system may include separate fluid systems for carrying out amplification and detection methods. Examples of integrated sequencing systems that can produce amplified nucleic acids and also determine the sequence of nucleic acids include, but are not limited to, the MiSeq® platform (Illumina, Inc., San Diego, CA) and the apparatus described in U.S. Patent Application No. 13 / 273,666, incorporated herein by reference.

[0173] The sequencing system described above sequences nucleic acid polymers present in a sample received by a sequencing device. As defined herein, “sample” and its derivatives are used in their broadest sense and include any subject, culture, etc., suspected to contain a target. In some embodiments, a sample includes DNA, RNA, PNA, LNA, chimeric or hybrid forms of nucleic acids. A sample may include any biological, clinical, surgical, agricultural, air, or water subject containing one or more nucleic acids. The term also includes any isolated nucleic acid sample, e.g., genomic DNA, fresh-frozen or formalin-fixed paraffin-embedded nucleic acid subject. A sample may also originate from a single individual, a collection of nucleic acid samples from genetically related members, nucleic acid samples from genetically unrelated members, nucleic acid samples from a single individual such as tumor and normal tissue samples (fitted), or a sample from a single source containing two different forms of genetic material, such as maternal and fetal DNA obtained from a maternal subject, or the presence of contaminating bacterial DNA in a sample containing plant or animal DNA. In some embodiments, the source of the nucleic acid material may include nucleic acids obtained from newborns, such as those typically used in newborn screening.

[0174] Nucleic acid samples may include high molecular weight substances such as genomic DNA (gDNA). Samples may include low molecular weight substances such as nucleic acid molecules obtained from FFPE or stored DNA samples. In another embodiment, the low molecular weight substance includes enzymatically or mechanically fragmented DNA. Samples may include cell-free circulating DNA. In some embodiments, samples may include nucleic acid molecules obtained from biopsies, tumors, scrapes, swabs, blood, mucus, urine, plasma, semen, hair, laser-captured microanatomy, surgical excisions, and other clinical or laboratory samples. In some embodiments, samples may be epidemiological, agricultural, forensic, or pathogenic samples. In some embodiments, samples may include nucleic acid molecules obtained from animals such as humans or mammalian sources. In another embodiment, samples may include nucleic acid molecules obtained from non-mammalian sources such as plants, bacteria, viruses, or fungi. In some embodiments, the source of nucleic acid molecules may be stored or extinct samples or species.

[0175] Furthermore, the methods and compositions disclosed herein may be useful for amplifying nucleic acid samples having low-quality nucleic acid molecules, such as degraded and / or fragmented genomic DNA, from forensic samples. In one embodiment, the forensic sample may include nucleic acids obtained from a crime scene, nucleic acids obtained from a missing persons DNA database, nucleic acids obtained from a laboratory associated with a forensic investigation, or forensic samples obtained by law enforcement agencies, one or more military services, or any such personnel. The nucleic acid sample may be crude DNA containing a purified sample or lysate derived from, for example, an oral swab, paper, cloth, or other substrate that can be impregnated with saliva, blood, or other bodily fluids. Thus, in some embodiments, the nucleic acid sample may include small amounts of DNA or fragmented portions of DNA, such as genomic DNA. In some embodiments, the target sequence may be present in one or more bodily fluids, including, but not limited to, blood, sputum, plasma, semen, urine, and serum. In some embodiments, the target sequence may be obtained from the victim's hair, skin, tissue sample, autopsy, or corpse. In some embodiments, nucleic acids containing one or more target sequences may be obtained from a deceased animal or human. In some embodiments, the target sequence may include nucleic acids obtained from non-human DNA, such as microbial, plant, or entomological DNA. In some embodiments, the target sequence or amplified target sequence is intended for human identification. In some embodiments, this disclosure generally relates to a method for characterizing forensic specimens. In some embodiments, this disclosure generally relates to a human identification method using one or more target-specific primers disclosed herein, or one or more target-specific primers designed using the primer design criteria outlined herein. In one embodiment, a forensic specimen or human identification specimen containing at least one target sequence may be amplified using one or more of the target-specific primers disclosed herein, or using the primer criteria outlined herein.

[0176] The components of the call recalibration system 106 may include software, hardware, or both. For example, the components of the call recalibration system 106 may include one or more instructions stored on a computer-readable storage medium and executable by a processor of one or more computing devices (e.g., client device 108). When executed by one or more processors, the computer-executable instructions of the call recalibration system 106 can cause the computing device to perform the bubble detection method described herein. Alternatively, the components of the call recalibration system 106 may include hardware, such as a dedicated processing unit for performing a particular function or set of functions. Additionally, or alternatively, the components of the call recalibration system 106 may include a combination of computer-executable instructions and hardware.

[0177] Furthermore, the components of the Call Recalibration System 106 that perform the functions described herein may be implemented, for example, as part of a standalone application, as a module of an application, as a plug-in to an application, as a library function that can be called by other applications, and / or as a cloud computing model. Thus, the components of the Call Recalibration System 106 may be implemented as part of a standalone application on a personal computing device or a mobile device. Additionally or alternatively, the components of the Call Recalibration System 106 may be implemented in any application that provides sequencing services, including, but not limited to, Illumina BaseSpace, Illumina DRAGEN, or Illumina TruSight software. "Illumina," "BaseSpace," "DRAGEN," and "TruSight" are registered trademarks or trademarks of Illumina, Inc. in the United States and / or other countries.

[0178] Embodiments of the present disclosure may include, or utilize, a dedicated or general-purpose computer, including, for example, one or more processors and system memory, computer hardware, as will be discussed in more detail below. Embodiments within the scope of the present disclosure also include physical and other computer-readable media for carrying or storing computer-executable instructions and / or data structures. In particular, one or more of the processes described herein may be embodied in a non-temporary computer-readable medium and at least partially implemented as instructions executable by one or more computing devices (e.g., any of the media content access devices described herein). Generally, a processor (e.g., a microprocessor) receives instructions from a non-temporary computer-readable medium (e.g., memory), executes those instructions, and thereby performs one or more processes, including one or more of the processes described herein.

[0179] A computer-readable medium can be any available medium that can be accessed by a general-purpose computer system or a dedicated computer system. A computer-readable medium that stores computer-executable instructions is a non-temporary computer-readable storage medium (device). A computer-readable medium that carries computer-executable instructions is a transmission medium. Thus, embodiments of the present disclosure may include, but are not limited to, two distinctly different types of computer-readable mediums: a non-temporary computer-readable storage medium (device) and a transmission medium.

[0180] Non-temporary computer-readable storage media (devices) include RAM, ROM, EEPROM, CD-ROM, solid-state drives (SSDs) (e.g., RAM-based), flash memory, phase-change memory (PCM), other types of memory, other optical disk storage, magnetic disk storage or other magnetic storage devices, or any other media that can be used to store desired program code means in the form of computer-executable instructions or data structures and can be accessed by a general-purpose or dedicated computer.

[0181] A “network” is defined as one or more data links that enable the transfer of electronic data between computer systems and / or modules and / or other electronic devices. When information is transferred to or provided to a computer via a network or another communication connection (either hardwired, wireless, or a combination of hardwired and wireless), the computer appropriately recognizes the connection as a transmission medium. A transmission medium can be used to carry desired program code means in the form of computer-executable instructions or data structures and may include networks and / or data links that can be accessed by general-purpose or dedicated computers. The above combinations should also be included within the scope of computer-readable media.

[0182] Furthermore, upon reaching various computer system components, program code in the form of computer-executable instructions or data structures can be automatically transferred from the transmission medium to a non-temporary computer-readable storage medium (device) (or vice versa). For example, computer-executable instructions or data structures received via a network or data link may be buffered in RAM within a network interface module (e.g., NIC) and then ultimately transferred to computer system RAM and / or a less volatile computer storage medium (device) within the computer system. Therefore, it should be understood that non-temporary computer-readable storage media (devices) can be included in computer system components that also (or more primarily) utilize the transmission medium.

[0183] Computer executable instructions include, for example, instructions and data that, when executed by a processor, cause a general-purpose computer, a dedicated computer, or a dedicated processing unit to perform a particular function or set of functions. In some embodiments, computer executable instructions are executed on a general-purpose computer and transform the general-purpose computer into a dedicated computer implementing the elements of the Disclosure. Computer executable instructions may be, for example, binary, intermediate format instructions such as assembly language, or even source code. While the subject matter is described in language specific to structural features and / or methodological behavior, it should be understood that the subject matter as defined in the appended claims is not necessarily limited to the described features or behaviors. Rather, the described features and behaviors are disclosed as exemplary forms that implement the claims.

[0184] Those skilled in the art will understand that the disclosure can be implemented in network computing environments having many types of computer system configurations, including personal computers, desktop computers, laptop computers, message processors, handheld devices, multiprocessor systems, microprocessor-based or programmable consumer electronics, network PCs, minicomputers, mainframe computers, mobile phones, PDAs, tablets, pagers, routers, switches, and the like. The disclosure can also be implemented in distributed system environments where both local and remote computer systems linked over a network (by hardwired data links, wireless data links, or a combination of hardwired and wireless data links) perform tasks. In a distributed system environment, program modules can reside in both local and remote memory storage devices.

[0185] Embodiments of this disclosure can also be implemented in a cloud computing environment. In this specification, “cloud computing” is defined as a model for enabling on-demand network access to a shared pool of configurable computing resources. For example, cloud computing may be adopted in the market to provide ubiquitous and convenient on-demand access to a shared pool of configurable computing resources. The shared pool of configurable computing resources can be rapidly configured via virtualization, exposed with low administrative effort or service provider interaction, and then scaled accordingly.

[0186] Cloud computing models can comprise a variety of characteristics, such as on-demand self-service, wide-area network access, resource pooling, rapid resilience, and measured service. Cloud computing models can also expose various service models, such as Software as a Service (SaaS), Platform as a Service (PaaS), and Infrastructure as a Service (IaaS). Cloud computing models can also be deployed using different deployment models, such as private clouds, community clouds, public clouds, and hybrid clouds. In this specification and in the claims, “cloud computing environment” refers to an environment in which cloud computing is employed.

[0187] Figure 10 shows a block diagram of a computing device 1000 that may be configured to perform one or more of the processes described above. It will be understood that one or more computing devices, such as computing device 1000, may implement the call recalibration system 106 and the sequence determination system 104. As shown in Figure 10, computing device 1000 may include a processor 1002, memory 1004, storage device 1006, I / O interface 1008, and communication interface 1010, which may be communicatively coupled by a communication infrastructure 1012. In certain embodiments, computing device 1000 may include fewer or more components than those shown in Figure 10. The following paragraphs describe in more detail the components of computing device 1000 shown in Figure 10.

[0188] In one or more embodiments, the processor 1002 includes hardware for executing instructions, such as instructions that constitute a computer program. For example, but not limited to, in order to execute instructions for dynamically modifying a workflow, the processor 1002 may retrieve (or fetch) instructions from internal registers, internal cache, memory 1004, or storage device 1006, decode them, and execute them. Memory 1004 may be volatile or non-volatile memory used to store data, metadata, and programs for execution by the processor. Storage device 1006 includes storage, such as a hard disk, flash disk drive, or other digital storage device, for storing data or instructions for carrying out the methods described herein.

[0189] The I / O interface 1008 allows a user to provide input to the computing device 1000 and receive output from it, and separately, to transfer data to it and receive data from it. The I / O interface 1008 may include a mouse, keypad or keyboard, touchscreen, camera, optical scanner, network interface, modem, other known I / O devices, or a combination of such I / O interfaces. The I / O interface 1008 may include, but is not limited to, one or more devices for presenting output to the user, including a graphics engine, a display (e.g., a display screen), one or more output drivers (e.g., display drivers), one or more audio speakers, and one or more audio drivers. In a particular embodiment, the I / O interface 1008 is configured to provide graphical data to the display for presentation to the user. The graphical data may represent one or more graphical user interfaces and / or any other graphical content that may be useful in a particular implementation.

[0190] The communication interface 1010 may include hardware, software, or both. In any case, the communication interface 1010 may provide one or more interfaces for communication (e.g., packet-based communication) between the computing device 1000 and one or more other computing devices or networks. For example, but not limited to, the communication interface 1010 may include a network interface controller (NIC) or network adapter for communication with Ethernet or other wired-based networks, or a wireless NIC (WNIC) or wireless adapter for communication with wireless networks such as Wi-Fi.

[0191] Additionally, the communication interface 1010 can facilitate communication with various types of wired or wireless networks. The communication interface 1010 can also facilitate communication using various communication protocols. The communication infrastructure 1012 may also include hardware, software, or both that connect the components of the computing device 1000 to one another. For example, the communication interface 1010 may enable multiple computing devices connected by a particular infrastructure to communicate with each other using one or more networks and / or protocols to carry out one or more aspects of the processes described herein. For example, a sequencing process may enable multiple devices (e.g., client devices, sequencing devices, and server devices) to exchange information such as sequencing data and error notifications.

[0192] In the aforementioned specification, the disclosure was described with reference to certain exemplary embodiments. Various embodiments and aspects of the disclosure are described with reference to the details considered herein, and the accompanying drawings illustrate various embodiments. The above description and drawings are illustrative of the disclosure and should not be construed as limiting the disclosure. Numerous specific details are described in order to provide a complete understanding of the various embodiments of the disclosure.

[0193] This disclosure may be embodied in other specific forms without departing from its spirit or essential features. The embodiments described herein should be considered in all respects to be illustrative and not limiting. For example, the methods described herein may be carried out using fewer or more steps / operations, or the steps / operations may be carried out in a different order. In addition, the steps / operations described herein may be repeated or carried out in parallel with each other, or in parallel with different occurrences of the same or similar steps / operations. Accordingly, the scope of this application is indicated by the appended claims rather than by the foregoing description. All changes included in the meaning of the claims and equivalents are encompassed within those scopes.

Claims

1. It is a system, At least one processor, A non-temporary computer-readable medium containing instructions, wherein, when the instructions are executed by the at least one processor, the system For the sample nucleotide sequence, access one or more sequencing data files containing data on nucleotide reads and genotype calls in genomic coordinates. Extract the sequencing metric for the nucleotide read or genotype call from the one or more sequencing data files. Using a call recalibration machine learning model and based on the sequencing metric, generate one or more variant call classifications that indicate the accuracy of the genotype calls in one or more sequencing data files. A system that generates a recalibrated sequencing data file containing updated genotype calls in the genomic coordinates for the sample nucleotide sequence, based on the one or more mutation call classifications.

2. The system according to claim 1, further comprising, when executed by the at least one processor, instructions causing the system to generate one or more variant call classifications without using a call generation model to simultaneously generate the genotype calls.

3. The system according to claim 1, wherein extracting the sequencing metrics for the genotype calls includes extracting one or more read-based sequencing metrics from the alignment data files of one or more sequencing data files.

4. The system according to claim 1, wherein extracting the sequencing metric includes extracting one or more read-based sequencing metrics or call model generation sequencing metrics for the genotype call from a genotype call data file of one or more sequencing data files, the genotype call data file containing the genotype call.

5. The system according to claim 4, further comprising, when executed by the at least one processor, instructions causing the system to access the genotype call data file by accessing a mutation call format (VCF) file or a genomic mutation call format (gVCF) file containing mutation calls and non-mutation calls.

6. The system according to claim 4, wherein the genotype call data file containing the genotype call is generated on a computing device running a hardware accelerator, and the recalibrated sequencing data file is generated using a general-purpose processing unit as the at least one processor of the system.

7. The system according to claim 6, wherein the hardware accelerator includes a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), and the recalibrated sequence determination data file is generated using one or more of a central processing unit (CPU) or a graphical processing unit (GPU).

8. When executed by the at least one processor, the system Access an additional genotype call data file generated by a different version of the call generation model than the one that generated the aforementioned genotype call data file. From the aforementioned additional genotype call data file, additional sequencing metrics for additional genotype calls in genomic coordinates for additional sample nucleotide sequences are extracted. Using the call recalibration machine learning model and based on the additional sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in the additional genotype call data file. The system according to claim 4, further comprising an instruction to generate an additional recalibrated sequencing data file, which includes updated additional genotype calls in the genomic coordinates for the additional sample nucleotide sequences, based on the one or more additional mutation call classifications.

9. The system according to claim 1, further comprising instructions, when executed by the at least one processor, causing the system to generate one or more variant call classifications by generating one or more of the following: a false positive probability that the genotype call is false positive, a genotype error probability that the genotype of the genotype call is incorrect, or a true positive probability that the genotype call is true positive.

10. When executed by the at least one processor, the system The aforementioned genome coordinates were identified as multi-allele genome coordinates, Using the call recalibration machine learning model, one or more mutation call classifications are generated, wherein the genotype call includes one or more of the following: a reference probability in the multi-allele genome coordinate system that includes a homozygous reference genotype; a zygosity error probability in the multi-allele genome coordinate system that includes a genotypic zygosity error; or a true positive mutation probability in the multi-allele genome coordinate system that constitutes a true positive mutation. The system according to claim 1, further comprising an instruction to determine the updated genotype call by determining the updated genotype call in the multi-allele genomic coordinates based on one or more of the reference probability, the conjugation error probability, or the true positive mutation probability.

11. When executed by the at least one processor, the system Based on the aforementioned one or more variant call classifications, one or more of the base call quality metric, genotype probability metric, genotype metric, genotype likelihood metric, or genotype quality metric for the genotype call are modified. The system according to claim 1, further comprising an instruction to generate the recalibrated sequencing data file, which includes the modified base call quality metric, the modified genotype probability metric, the modified genotype metric, the modified genotype likelihood metric, or the modified genotype quality metric.

12. When executed by the at least one processor, the system Instead of determining the heterozygous mutant genotype call or homozygous mutant genotype call reported in one or more sequencing data files, determine the homozygous reference genotype call in the genomic coordinates. Instead of determining the homozygous reference genotype call or the homozygous mutant genotype call reported in one or more sequencing data files, determine the heterozygous mutant genotype call in the genomic coordinates, or The system according to claim 1, further comprising an instruction to generate the updated genotype call in the biallelic genomic coordinates for the sample nucleotide sequence as part of the recalibrated sequencing data file by determining the homozygous mutant genotype call in the genomic coordinates instead of the heterozygous mutant genotype call or the homozygous reference genotype call reported in the one or more sequencing data files.

13. When executed by the at least one processor, the system From the one or more sequencing data files, the sequencing metric for additional genotype calls at additional genomic coordinates for the sample nucleotide sequence is extracted. Using the call recalibration machine learning model and based on the sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in one or more sequencing data files. Based on the one or more additional variant call classifications, the base call quality metric for the additional genotype calls is modified to generate a modified base call quality metric that falls below the base call quality threshold. The system according to claim 1, further comprising instructions to annotate the additional genotype calls to indicate that the modified base call quality metric falls below the base call quality threshold.

14. When executed by the at least one processor, the system From the one or more sequencing data files, the sequencing metric for additional genotype calls at additional genomic coordinates for the sample nucleotide sequence is extracted. Using the call recalibration machine learning model and based on the sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in one or more sequencing data files. The system according to claim 1, further comprising an instruction to confirm the genotype call in the additional genomic coordinates for the sample nucleotide sequence based on the one or more additional mutation call classifications.

15. When executed by the at least one processor, the system Generate multiple recalibrated sequencing data files from multiple sequencing data files corresponding to multiple known genomes. The updated genotype calls from the aforementioned multiple recalibrated sequencing data files are compared with known mutations in the aforementioned multiple known genomes. The system according to claim 1, further comprising instructions to train the call recalibration machine learning model by adjusting the parameters of the call recalibration machine learning model based on the difference between the updated genotype call and the known mutation.

16. It is a system, At least one processor, A non-temporary computer-readable medium storing instructions, wherein when an instruction is executed by the at least one processor, the system For the sample nucleotide sequence, access one or more sequencing data files containing genotype calls in genomic coordinates. Extract the sequence determination metric for the genotype call from the one or more sequence determination data files mentioned above. Using a call recalibration machine learning model and based on the sequencing metric, generate one or more variant call classifications that indicate the accuracy of the genotype calls in one or more sequencing data files. A system that generates a recalibrated sequencing data file containing updated genotype calls in the genomic coordinates for the sample nucleotide sequence, based on the one or more mutation call classifications.

17. When executed by the at least one processor, the system Accessing the alignment data files of one or more sequencing data files, wherein the alignment data files include nucleotide reads corresponding to the genomic coordinates for the sample nucleotide sequence. The system according to claim 16, further comprising instructions to extract one or more read-based sequencing metrics of the sequencing metrics from the alignment data file, wherein the one or more read-based sequencing metrics correspond to the nucleotide reads.

18. The system according to claim 16, further comprising, when executed by the at least one processor, instructions causing the system to generate one or more variant call classifications without using a call generation model to simultaneously generate the genotype calls.

19. The system according to claim 16, wherein extracting the sequencing metric includes extracting one or more read-based sequencing metrics or call model generation sequencing metrics for the genotype call from a genotype call data file of one or more sequencing data files, the genotype call data file containing the genotype call.

20. The system according to claim 19, further comprising, when executed by the at least one processor, instructions causing the system to access the genotype call data file by accessing a mutation call format (VCF) or genomic mutation call format (gVCF) file containing mutation calls and non-mutation calls.

21. The system according to claim 19, wherein the genotype call data file containing the genotype call is generated on a computing device running a hardware accelerator, and the recalibrated sequencing data file is generated using a general-purpose processing unit as the at least one processor of the system.

22. The system according to claim 21, wherein the hardware accelerator includes a field-programmable gate array (FPGA) or an application-specific integrated circuit (ASIC), and the recalibrated sequence determination data file is generated using one or more of a central processing unit (CPU) or a graphical processing unit (GPU).

23. When executed by the at least one processor, the system Access an additional genotype call data file generated by a different version of the call generation model than the one that generated the aforementioned genotype call data file. From the aforementioned additional genotype call data file, additional sequencing metrics for additional genotype calls in genomic coordinates for additional sample nucleotide sequences are extracted. Using the call recalibration machine learning model and based on the additional sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in the additional genotype call data file. The system according to claim 19, further comprising an instruction to generate an additional recalibrated sequencing data file, which includes updated additional genotype calls in the genomic coordinates for the additional sample nucleotide sequences, based on the one or more additional mutation call classifications.

24. The system according to claim 16, further comprising an instruction, when executed by the at least one processor, causing the system to generate one or more variant call classifications by generating one or more of the following: a false positive probability that the genotype call is false positive, a genotype error probability that the genotype of the genotype call is incorrect, or a true positive probability that the genotype call is true positive.

25. When executed by the at least one processor, the system The aforementioned genome coordinates were identified as multi-allele genome coordinates, Using the call recalibration machine learning model, one or more mutation call classifications are generated, wherein the genotype call includes one or more of the following: a reference probability in the multi-allele genome coordinate system that includes a homozygous reference genotype; a zygosity error probability in the multi-allele genome coordinate system that includes a genotypic zygosity error; or a true positive mutation probability in the multi-allele genome coordinate system that constitutes a true positive mutation. The system according to claim 16, further comprising an instruction to determine the updated genotype call by determining the updated genotype call in the multi-allele genomic coordinates based on one or more of the reference probability, the conjugation error probability, or the true positive mutation probability.

26. When executed by the at least one processor, the system Based on the aforementioned one or more variant call classifications, one or more of the base call quality metric, genotype probability metric, genotype metric, genotype likelihood metric, or genotype quality metric for the genotype call are modified. The system according to claim 16, further comprising an instruction to generate the recalibrated sequencing data file, which includes the modified base call quality metric, the modified genotype probability metric, the modified genotype metric, the modified genotype likelihood metric, or the modified genotype quality metric.

27. When executed by the at least one processor, the system Instead of determining the heterozygous mutant genotype call or homozygous mutant genotype call reported in one or more sequencing data files, determine the homozygous reference genotype call in the genomic coordinates. Instead of determining the homozygous reference genotype call or the homozygous mutant genotype call reported in one or more sequencing data files, determine the heterozygous mutant genotype call in the genomic coordinates, or The system according to claim 16, further comprising an instruction to generate the updated genotype call in the biallelic genomic coordinates for the sample nucleotide sequence as part of the recalibrated sequencing data file by determining the homozygous variant genotype call in the genomic coordinates instead of the heterozygous variant genotype call or the homozygous reference genotype call reported in one or more sequencing data files.

28. When executed by the at least one processor, the system From the one or more sequencing data files, the sequencing metric for additional genotype calls at additional genomic coordinates for the sample nucleotide sequence is extracted. Using the call recalibration machine learning model and based on the sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in one or more sequencing data files. Based on the one or more additional variant call classifications, the base call quality metric for the additional genotype calls is modified to generate a modified base call quality metric that falls below the base call quality threshold. The system according to claim 16, further comprising instructions to annotate the additional genotype calls to indicate that the modified base call quality metric falls below the base call quality threshold.

29. When executed by the at least one processor, the system From the one or more sequencing data files, the sequencing metric for additional genotype calls at additional genomic coordinates for the sample nucleotide sequence is extracted. Using the call recalibration machine learning model and based on the sequencing metric for the additional genotype calls, generate one or more additional variant call classifications indicating the accuracy of the additional genotype calls in one or more sequencing data files. The system according to claim 16, further comprising an instruction to confirm the genotype call in the additional genomic coordinates for the sample nucleotide sequence based on the one or more additional mutation call classifications.

30. When executed by the at least one processor, the system Generate multiple recalibrated sequencing data files from multiple sequencing data files corresponding to multiple known genomes. The updated genotype calls from the aforementioned multiple recalibrated sequencing data files are compared with known mutations in the aforementioned multiple known genomes. The system according to claim 16, further comprising instructions to train the call recalibration machine learning model by adjusting the parameters of the call recalibration machine learning model based on the difference between the updated genotype call and the known mutation.

31. When executed by the at least one processor, the system Using the call recalibration machine learning model trained on indel training data, generate one or more variant call classifications for one or more candidate insertions or deletions (indels). The system according to claim 16, further comprising an instruction to generate the updated genotype call indicating the presence or absence of the indel in the genomic coordinates for the sample nucleotide sequence, based on the one or more mutation call classifications for the one or more candidate indels.

32. When executed by the at least one processor, the system Using the call recalibration machine learning model trained on single nucleotide variant (SNV) training data, generate one or more variant call classifications for one or more candidate SNVs. The system according to claim 16, further comprising an instruction to generate the updated genotype call indicating the presence or absence of the SNV in the genomic coordinates for the sample nucleotide sequence, based on the one or more mutation call classifications for the one or more candidate SNVs.

Images