Tumor whole exon sequencing panoramic variation map analysis method and system based on Argo Workflows

Abstract

The invention discloses a tumor whole exon sequencing panoramic variation map analysis method and system based on Argo Workflows. The method comprises the steps that all analysis tools involved in the variation type analysis process of whole exon sequencing original data of a tumor sample and a normal sample paired with the tumor sample are packaged into independent Docker containers through Argo Workflows; performing sequence alignment on the whole exon sequencing original data by using two parallel task branches; performing data preprocessing by using two parallel task branches; s4, performing parallel execution of embryonic line mutation analysis, copy number variation analysis, MSI state analysis and somatic mutation detection on the tumor and normal sample sequence comparison data; calculating a tumor neoantigen prediction result and a TMB value; and summarizing to generate a tumor whole exon sequencing panoramic variation map. The method simplifies operation, reduces manual intervention, improves accuracy, improves analysis efficiency and has repeatability.

Description

Technical Field

[0001] This invention pertains to bioinformatics analysis technology for tumor gene sequencing data, and particularly relates to a panoramic variation map analysis method based on ArgoWorkflows whole-exome sequencing of tumors, as well as a system using this analysis method. Background Technology

[0002] The occurrence and development of tumors is a complex, multi-stage process, with genomic instability and the resulting series of gene variations being the core driving factors. Whole exome sequencing (WES), as a cost-effective next-generation sequencing technology, can capture approximately 1-2% of the coding regions (exons) and some adjacent intronic regions in the genome. These regions are rich in gene variation information closely related to disease occurrence, development, and drug response. Therefore, tumor whole exome sequencing (WES) has been widely used in clinical and research fields such as precision diagnosis, prognostic assessment, targeted therapy guidance, and the exploration of novel biomarkers for tumors.

[0003] Whole-exome sequencing (WES) of tumor samples and their paired normal samples (such as peripheral blood) can theoretically reveal a comprehensive map of tumor genome variations. These variations are diverse and mainly include:

[0004] Somatic single nucleotide variants (sSNVs) and short insertions / deletions (sIndels): These are key mutations that drive tumorigenesis and development, and are also the sites of action for many targeted drugs.

[0005] Copy number variations (CNVs) refer to increases (amplifications) or decreases (deletions) in the copy number of large segments of the genome. They are commonly seen in homozygous deletions of tumor suppressor genes or amplifications of oncogenes and have a significant impact on tumor phenotypes.

[0006] Structural variations (SVs) include large-segment translocations, inversions, duplications, and complex rearrangements, which may lead to gene fusions, gene expression disorders, or loss of function.

[0007] Microsatellite instability (MSI) is an important biomarker for certain tumor types (such as colorectal cancer and endometrial cancer) and a predictor of the efficacy of immune checkpoint inhibitors, resulting from defects in the function of mismatch repair genes that alter the length of microsatellite sequences in the genome.

[0008] Tumor mutational burden (TMB) refers to the total number of somatic nonsynonymous mutations per unit length (usually per megabase) within the coding region of a tumor genome. A high TMB usually indicates a better response to immunotherapy.

[0009] Germline single nucleotide variants (SNVs) and short insertions / deletions (IGDs): These genetic variants may be associated with tumor susceptibility, drug metabolism, and certain genetic syndromes.

[0010] Obtaining comprehensive and accurate variant results is crucial for developing individualized cancer treatment plans, assessing genetic risks, and advancing precision medicine. However, extracting this multidimensional variant information from the massive amounts of raw sequencing data (typically tens to hundreds of gigabytes) generated by whole-exome sequencing (WES) presents significant bioinformatics challenges.

[0011] Current workflows for analyzing whole-exome sequencing (WES) data in tumors typically rely on a combination of independent bioinformatics tools and algorithms tailored to specific variant types. For example:

[0012] Sequence alignment: Tools such as BWA (Burrows-Wheeler Aligner) or Bowtie2 are typically used to align sequencing reads to a reference genome.

[0013] Somatic SNV / Indel detection: Tools such as Mutect2 (GATK), VarScan2, and Strelka2 are widely used. They identify somatic mutations by comparing the results of tumor and normal samples, and require complex filtering strategies to remove false positives.

[0014] CNV detection requires specialized algorithms, such as CNVkit. These tools infer copy number changes based on information such as read depth and B allele frequency (BAF).

[0015] SV detection: Detecting SVs from whole exome sequencing (WES) data is more challenging, and commonly used tools include Manta.

[0016] MSI status assessment: Tools such as MSIsensor are typically used to determine MSI status by analyzing the differences in read distribution at specific microsatellite sites between tumor and normal samples.

[0017] TMB calculation: After identifying sSNV / sIndel, TMB is estimated by counting nonsynonymous mutations and correcting for exon size.

[0018] Germline variation detection: Tools such as GATK HaplotypeCaller are commonly used to detect variations in normal samples.

[0019] Neoantigen testing: Tools such as PVACtools are commonly used for testing.

[0020] The above "divide and conquer" analytical strategies have the following drawbacks:

[0021] (1) Complex process and cumbersome operation: Researchers or clinical analysts need to learn, configure, run and maintain multiple independent software tools, each with its own unique parameter settings and input / output formats, making the entire analysis process lengthy and difficult to standardize.

[0022] (2) Difficult data integration and poor result consistency: Different tools produce mutation results in different formats, requiring additional steps for format conversion and integration. More importantly, the internal logic and optimization objectives of different algorithms differ, which may lead to potential inconsistencies or even conflicts between the analysis results of different mutation types on the same dataset.

[0023] (3) High computational resource consumption and high time cost: Running multiple independent analysis modules usually requires a lot of computational resources (CPU, memory, storage), and the serial or partially parallel analysis steps lengthen the overall analysis cycle.

[0024] (4) High requirements for bioinformatics skills: Users not only need to understand the biological significance of various types of variants, but also need to have a deep background in bioinformatics in order to effectively select appropriate tools, optimize parameters and interpret results. This limits the widespread application of software tools in clinical practice.

[0025] Given the aforementioned limitations of existing technologies—namely, the reliance of tumor whole-exome sequencing (WES) data analysis workflows on multiple independent and heterogeneous bioinformatics tools, which leads to complex, time-consuming, and difficult data integration processes, high skill requirements, and difficulty in achieving coordinated, efficient, and accurate comprehensive detection and interpretation of various key variant types such as somatic SNV / Indel, CNV, SV, MSI, TMB, neoantigens, and germline SNV / Indel—this severely restricts the ability of tumor whole-exome sequencing (WES) data to rapidly generate panoramic variant atlases and guide precision treatment in clinical practice. Summary of the Invention

[0026] The first objective of this invention is to provide an automated, convenient, accurate, and efficient method for comprehensive and reproducible whole-exome sequencing-based panoramic mutation atlas analysis of tumors, which is based on Argo Workflows.

[0027] The first objective of this invention is achieved through the following technical measures: a panoramic variant mapping method for tumor whole-exome sequencing based on Argo Workflows, characterized by comprising the following steps:

[0028] S1. Using Argo Workflows, all the analysis tools involved in the variation type analysis process of whole exome sequencing (WES) raw data of tumor samples and their paired normal samples are packaged into independent Docker containers, and each container is defined as an independent task.

[0029] S2. Use two parallel task branches to perform sequence alignment on the raw whole exome sequencing (WES) data: First, align the sequencing reads of the tumor sample to the reference genome to generate sequence alignment data for the tumor sample; Second, align the sequencing reads of the normal sample paired with the tumor sample to the same reference genome to generate sequence alignment data for the normal sample.

[0030] S3. Use two parallel task branches to preprocess the data obtained from step S2, that is, the two task branches perform base quality score recorrection on the tumor sample sequence alignment data and the normal sample sequence alignment data, respectively.

[0031] S4. Perform the following analysis and processing workflow in parallel on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data:

[0032] (1) Perform germline mutation analysis on the sequence alignment data of normal samples after data preprocessing to obtain the germline mutation analysis results;

[0033] (2) Copy number variation analysis was performed on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the copy number variation analysis results;

[0034] (3) Perform MSI status analysis on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the MSI status analysis results;

[0035] (4) Perform somatic mutation detection on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data;

[0036] S5. Based on the sequence alignment data of tumor samples after somatic mutation detection, the sequence alignment data of normal samples, and the patient's HLA typing information, predict tumor neoantigens and obtain tumor neoantigen prediction results.

[0037] S6. Calculate the TMB value by comparing the sequence alignment data of tumor samples and normal samples after somatic mutation detection.

[0038] S7. The results of somatic mutation analysis, germline mutation analysis, MSI status analysis, copy number variation analysis, TMB value and tumor neoantigen prediction are summarized to generate a panoramic mutation map of tumor whole exome sequencing.

[0039] This invention transforms a complex, multi-variant detection process (somatic SNV / Indel, CNV, SV, MSI, TMB, germline SNV / Indel, neoantigens) into a declarative, containerized, and automated orchestration and execution using Argo Workflows. Leveraging Argo Workflows' features such as DAG definition, containerized tasks, input / output artifact management, parallel execution, error retries, and resource management, it establishes a method and architecture for an end-to-end, reproducible, and scalable comprehensive tumor whole-exome sequencing (WES) analysis system. This simplifies operation, reduces manual intervention, improves accuracy, enhances analysis efficiency, and ensures reproducibility, enabling comprehensive and in-depth analysis of tumor whole-exome sequencing (WES) data.

[0040] In step S2 of this invention, each task branch first uses the BWA-MEM tool to perform sequence alignment to generate a SAM file, and then uses the Samtools tool to convert the format of the SAM file, sort it, and index it to generate a BAM file.

[0041] In step S3 of this invention, each task branch uses the BaseRecalibrator tool to perform base quality score recalibration and outputs a BAM file.

[0042] In step S4(1) of this invention, germline SNV and Indel detection are first performed using the Germline HaplotypeCaller tool to generate a VCF file. Then, the VCF2MAF tool is used to convert the VCF file into an MAF file for easy integration and interpretation. Finally, the filtered germline mutation analysis results are input into the results summary.

[0043] In step S4⑵ of this invention, the CNVkit tool is used to perform copy number variation analysis, and then the filtered copy number variation analysis input results are summarized.

[0044] In step S4⑶ of this invention, the MSIsensor2 tool is used to perform MSI status analysis.

[0045] In step S4(4) of this invention, multiple tools are used in parallel to detect somatic cell SNVs / Indels and SVs. Specifically, the Mutect2 tool is used to detect somatic cell SNVs and Indels, generating a VCF file; the Samtools mpileup tool is used in parallel to generate a pileup format file, which is then used again with the VarScan2 tool to detect somatic cell SNVs and Indels, generating a VCF file; the Manta tool is used in parallel to detect SVs, then the SVs are annotated, fusion gene analysis is performed, and the results are summarized; simultaneously, the Manta tool is used to detect SVs and generate candidate Indel regions, which are then input into the Strelka2 tool for somatic cell SNV and Indel detection, generating a VCF file; finally, the VCF files output by the Mutect2, VarScan2, and Strelka2 tools are converted into MAF files in parallel.

[0046] In step S5 of this invention, the tumor sample sequence alignment data and normal sample sequence alignment data after somatic mutation detection refer to MAF files output and converted by the Mutect2, VarScan2 and Strelka2 tools, and the tumor neoantigen is predicted using the PVACTools tool.

[0047] In step S6 of this invention, the sequence alignment data of tumor samples and normal samples detected by somatic mutation specifically refers to the MAF file generated by converting the detection results output by the Mutect2 tool, the MAF file generated by converting the somatic single nucleotide variant (SNV) detection results output by the VarScan2 tool, and the MAF file generated by converting the somatic SNV detection results output by the Strelka2 tool. Based on the above MAF files, the calculation of tumor mutation burden (TMB) can be performed in parallel. At the same time, the filtered somatic detection results are directly generated into an HTML file that allows users to directly view the mutation results. The MAF files converted from the detection data output by the Mutect2 tool, the MAF files converted from the detection data output by the VarScan2 tool, and the MAF files converted from the detection data output by the Strelka2 tool are respectively input into the result summary.

[0048] The second objective of this invention is to provide a system that uses the above-described Argo Workflows-based panoramic variant mapping method for tumor whole-exome sequencing.

[0049] The second objective of this invention is achieved through the following technical measures: a system using the above-mentioned ArgoWorkflows-based panoramic variant mapping analysis method for whole-exome sequencing of tumors, characterized by comprising:

[0050] The sequence alignment module is used to perform sequence alignment processing on the raw whole exome sequencing (WES) data of tumor samples and their paired normal samples to obtain tumor sample sequence alignment data and normal sample sequence alignment data.

[0051] The data preprocessing module is used to perform base quality score recorrection on tumor sample sequence alignment data and normal sample sequence alignment data;

[0052] The germline mutation analysis module is used to perform germline mutation analysis on the preprocessed normal sample sequence alignment data.

[0053] The copy number variation analysis module is used to perform copy number variation analysis on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data.

[0054] The MSI status analysis module is used to perform MSI status analysis on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data.

[0055] The somatic mutation detection module is used to detect somatic mutations in preprocessed tumor sample sequence alignment data and normal sample sequence alignment data.

[0056] The tumor neoantigen analysis module is used to predict tumor neoantigens based on sequence alignment data of tumor samples after somatic mutation detection, sequence alignment data of normal samples, and patient HLA typing information.

[0057] The TMB analysis module is used to calculate TMB on the sequence alignment data of tumor samples and normal sample sequences that have undergone somatic mutation detection.

[0058] The results summary module is used to summarize the results of somatic mutation data, germline mutation analysis, MSI status analysis, copy number variation analysis, TMB value and tumor neoantigen prediction to generate a panoramic mutation map of tumor whole exome sequencing.

[0059] The input of the sequence alignment module is used to receive raw whole-exome sequencing (WES) data of tumor samples and their paired normal samples. The output of the sequence alignment module is connected to the input of the data preprocessing module. The output of the data preprocessing module is connected to the input of the germline mutation analysis module, the MSI status analysis module, the copy number variation analysis module, and the somatic mutation detection module, respectively. The outputs of the germline mutation analysis module, the copy number variation analysis module, and the MSI status analysis module are connected to the result aggregation module. The output of the somatic mutation detection module is connected to the input of the tumor neoantigen analysis module and the TMB analysis module, respectively. The outputs of the tumor neoantigen analysis module and the TMB analysis module are connected to the result aggregation module, respectively.

[0060] Compared with the prior art, the present invention has the following significant effects:

[0061] (1) Significantly improves analytical efficiency and throughput: This invention, through the automated orchestration and parallel processing capabilities of Argo Workflows, can automate and parallelize multiple analytical steps that originally required manual sequential execution, significantly shortening the total turnaround time (TAT) from raw sequencing data to obtaining comprehensive variant results. For clinical testing laboratories or research institutions that need to process large numbers of samples, this invention can significantly improve analytical throughput and meet the demand for high timeliness.

[0062] (2) Enhancing the comprehensiveness and accuracy of analysis results: This invention can systematically detect and integrate multiple key variant types, including somatic SNV / Indel, CNV, SV, MSI status, TMB value, neoantigen prediction, and germline SNV / Indel, within a unified workflow, providing a panoramic view of the tumor genome and avoiding the omission of important clinical information due to selective analysis or tool limitations. By integrating the results of multiple somatic SNV / Indel detection tools and combining them with standardized preprocessing and quality control, the robustness and accuracy of variant detection can be improved, reducing the bias or false negatives / false positives that may be introduced by a single tool.

[0063] (3) Significantly reduced operational complexity and technical barriers: The "one-stop" design of this invention encapsulates the complex bioinformatics analysis process. Users do not need to delve into the details and parameter configurations of each tool; they only need to provide standard input data to start the analysis and obtain the final results, greatly reducing the requirements for users' professional bioinformatics skills. This enables clinicians or researchers without a strong bioinformatics background to easily utilize whole-exome sequencing (WES) data for in-depth analysis, promoting the widespread application of genomic information in clinical practice.

[0064] (4) Enhancing the Standardization and Reproducibility of the Analysis Process: Based on Argo Workflows and containerization technology, the analysis process of this invention features a high degree of standardization and reproducibility. The same input data consistently produces consistent results under the same process, which is crucial for quality control in clinical diagnosis and validation of research findings. The definition and versions of the process can be effectively managed, facilitating tracking and auditing.

[0065] (5) Facilitating clinical decision-making and personalized treatment: By rapidly and accurately providing a panoramic atlas of tumor mutations, including known drug targets, drug resistance mutations, immunotherapy-related biomarkers (MSI, TMB, neoantigens), genetic susceptibility, and other information, this invention can provide clinicians with strong and comprehensive data support for developing personalized treatment plans.

[0066] (6) Optimize computing resource utilization and cost-effectiveness: Argo Workflows can effectively schedule and manage computing resources on Kubernetes clusters, dynamically allocating resources according to task requirements and avoiding resource waste. Automated processes reduce the time cost of manual operations and the rework cost caused by potential errors, improving overall cost-effectiveness.

[0067] (7) Enhanced Research and Development Capabilities: The modular design allows researchers to easily replace or add new analytical tools or algorithm modules to evaluate the performance of different analytical strategies or quickly integrate the latest research findings. Comprehensive variant data provides a rich data foundation for exploring new biomarkers, understanding the mechanisms of tumor development and progression, and developing new treatment strategies. Attached Figure Description

[0068] The present invention will now be described in further detail with reference to the accompanying drawings and specific embodiments.

[0069] Figure 1 This is a schematic diagram of the process and system composition of the method of the present invention. Detailed Implementation

[0070] The present invention will now be described in detail with reference to the embodiments and accompanying drawings to help those skilled in the art better understand the inventive concept of the present invention. However, the scope of protection of the claims of the present invention is not limited to the following embodiments. For those skilled in the art, all other embodiments obtained without creative effort without departing from the inventive concept of the present invention are within the scope of protection of the present invention.

[0071] This invention utilizes Argo Workflows for containerized encapsulation and process orchestration. Argo Workflows is an open-source, Kubernetes-based container-native workflow engine for orchestrating parallel jobs.

[0072] like Figure 1 As shown, this invention discloses a method for constructing and analyzing a panoramic variant map of tumors based on Argo Workflows whole-exome sequencing, including the following steps:

[0073] S1. Using Argo Workflows, each tool used in the various analysis processes for variant type analysis of raw whole exome sequencing (WES) data from tumor samples and their paired normal samples is packaged into an independent Docker container and defined as a task.

[0074] Figure 1 Each independent analysis step or tool (such as BWA-MEM, Samtools, BaseRecalibrator, Mutect2, CNVkit, etc.) is encapsulated as a separate Docker container and defined as a step or task within Argo Workflows. The relationships between these steps (such as...) Figure 1 (As indicated by the middle arrow) This is explicitly stated in the Argo Workflows YAML definition file, ensuring the correctness of the data flow and the orderly execution of the analysis.

[0075] S2. Use two parallel task branches to perform sequence alignment processing on the raw whole exome sequencing (WES) data. One task branch aligns the sequencing reads of the tumor sample to the reference genome to generate tumor sample sequence alignment data, and the other task branch aligns the sequencing reads of the normal sample paired with the tumor sample to the reference genome to generate normal sample sequence alignment data.

[0076] This step is defined in Argo Workflows as two parallel task branches (one for Normal and one for Tumor), each branch containing serial sub-steps of BWA-MEM and Samtools.

[0077] The FASTQ files of Normal and Tumor samples were aligned using BWA-MEM. The resulting SAM files were then converted (SAM to BAM), sorted, and indexed using Samtools to generate the BAM files required for subsequent analysis.

[0078] S3. Use two parallel task branches to preprocess the data obtained from step S2, that is, the two task branches perform base quality score recorrection on the tumor sample sequence alignment data and the normal sample sequence alignment data, respectively.

[0079] This step receives the BAM file from step S2 as input and is also divided into two parallel tasks: Normal and Tumor. The output BAM file processed by BQSR (hereinafter referred to as "analysis-ready BAM file") serves as the common input for multiple subsequent variant detection modules.

[0080] Base quality score recalibration (BQSR) is performed on the BAM files of Normal and Tumor samples using BaseRecalibrator (usually a tool in the GATK suite), and the BAM files are output.

[0081] S4. Perform the following analysis procedures in parallel on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data:

[0082] (1) Perform germline mutation analysis on the sequence alignment data of normal samples after data preprocessing to obtain the germline mutation analysis results;

[0083] This step relies on the analysis-ready BAM file of the Normal sample, whose output Germline MAF file will be imported into the final "Results Summary" step.

[0084] Germline HaplotypeCaller (usually GATK) is used to perform germline SNV and Indel detection on the analysis-ready BAM files of normal samples to generate VCF files. Then, the VCF2MAF tool is used to convert the germline VCF files to MAF (Mutation Annotation Format) format for easy integration and interpretation. Finally, the filtered germline mutation analysis results are entered into the results summary.

[0085] (2) Copy number variation analysis was performed on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the copy number variation analysis results;

[0086] This step relies on the analysis-ready BAM file, whose output CNV results are incorporated into the final "Results Summary" step.

[0087] The CNVkit was used to perform copy number variation (CNV) analysis on paired Tumor and Normal analysis-ready BAM files, and the filtered copy number variation analysis results were then summarized.

[0088] (3) Perform MSI status analysis on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the MSI status analysis results;

[0089] This step relies on the Analysis Ready BAM file, whose output MSI status report is imported into the final "Results Summary" step.

[0090] MSI status analysis was performed on paired Tumor and Normal analysis-ready BAM files using MSIsensor2.

[0091] (4) Perform somatic mutation detection on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data;

[0092] This step employs multiple tools to detect somatic cell SNV / Indel and SV in parallel to improve the detection rate and the reliability of the results.

[0093] Somatic SNV and Indel detection were performed on paired Tumor and Normal analysis-ready BAM files using Mutect2 (GATK) to generate VCF files.

[0094] Generate a pileup format file using Samtools mpileup, then use VarScan2 to perform somatic cell SNV and Indel detection to generate a VCF file.

[0095] Structural variation (SV) detection was performed on paired Tumor and Normal analysis-ready BAM files using Manta, followed by SV annotation. After fusion gene analysis, the results were summarized and entered into the database. Simultaneously, SV detection was performed using Manta to generate candidate Indel regions, which were then entered into Strelka2 for somatic SNV and Indel detection to generate VCF files.

[0096] The four tasks above are executed in parallel, all taking the analysis-ready BAM files from Tumor and Normal as input. Their VCF outputs will either be used for subsequent MAF transformations or directly used for result summarization.

[0097] MAF conversion:

[0098] Define a MAF conversion task for each upstream VCF output, and these tasks are executed in parallel.

[0099] VarScan2 Indels MAF (VarScan2 Indel VCF -> MAF)

[0100] VarScan2 SNVs MAF (VarScan2 SNV VCF -> MAF)

[0101] Mutect2 VCF2MAF (Mutect2 VCF -> MAF)

[0102] Strelka2 Indels MAF (Strelka2 Indel VCF -> MAF)

[0103] Strelka2 SNVs MAF (Strelka2 SNV VCF -> MAF)

[0104] VCF outputs from different somatic SNV / Indel detection tools (Mutect2, VarScan2, Strelka2) are uniformly converted into MAF format to facilitate downstream TMB calculation, neoantigen prediction, and result integration.

[0105] S5. Based on the sequence alignment data of tumor samples after somatic mutation detection, the sequence alignment data of normal samples, and the patient's HLA typing information, predict tumor neoantigens and obtain tumor neoantigen prediction results.

[0106] This step relies on the somatic MAF file and HLA typing results output by S4. The list of neoantigens output will be incorporated into the final "Results Summary" step.

[0107] Using PVACTools (such as the pVACseq workflow), combine somatic mutation MAF files (usually selecting one or more high-quality MAF sources, or integrated MAFs) with patient HLA typing information to predict tumor neoantigens.

[0108] S6. Calculate the TMB value by comparing the sequence alignment data of tumor samples and normal samples after somatic mutation detection.

[0109] Define a TMB calculation task for each MAF input. These different TMB values ​​will be processed during the results aggregation phase.

[0110] Tumor mutation burden (TMB) was calculated based on somatic mutation data in MAF format, with TMB values ​​calculated based on VarScan2MAF, Mutect2MAF, and Strelka2MAF, respectively.

[0111] The filtered somatic cell test results will be directly generated into an HTML file of IGV results using the IGV report. Users can open the file to view the mutation results, and the somatic cell mutation results will also be copied into the final results.

[0112] S7. The results of somatic mutation analysis, germline mutation analysis, MSI status analysis, copy number variation analysis, TMB value and tumor neoantigen prediction are summarized to generate a panoramic mutation map of tumor whole exome sequencing.

[0113] This step is defined as the convergence point of Argo Workflows and depends on the completion of all upstream analysis steps. It includes the final results file compilation.

[0114] This step involves collecting, integrating, annotating, and formatting data from all upstream analysis steps (germline MAF, MSI analysis results, CNV analysis results, Manta's SV results VCF, somatic MAF from various tools, TMB values ​​from various tools, and neoantigen prediction results). Annotation provides comprehensive functional annotation for all variants (SNV / Indel, SV, CNV).

[0115] A system using the aforementioned Argo Workflows-based panoramic variant mapping method for tumor whole-exome sequencing includes:

[0116] The sequence alignment module is used to perform sequence alignment processing on the raw whole exome sequencing (WES) data of tumor samples and their paired normal samples to obtain tumor sample sequence alignment data and normal sample sequence alignment data.

[0117] The data preprocessing module is used to perform base quality score recorrection on tumor sample sequence alignment data and normal sample sequence alignment data;

[0118] The germline mutation analysis module is used to perform germline mutation analysis on the preprocessed normal sample sequence alignment data.

[0119] The copy number variation analysis module is used to perform copy number variation analysis on preprocessed tumor samples and paired normal sample sequence alignment data.

[0120] The MSI status analysis module is used to perform MSI status analysis on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data.

[0121] The somatic mutation detection module is used to detect somatic mutations in preprocessed tumor sample sequence alignment data and normal sample sequence alignment data.

[0122] The tumor neoantigen analysis module is used to predict tumor neoantigens based on sequence alignment data of tumor samples after somatic mutation detection, sequence alignment data of normal samples, and patient HLA typing information.

[0123] The TMB analysis module is used to calculate TMB on the sequence alignment data of tumor samples and normal sample sequences that have undergone somatic mutation detection.

[0124] The results summary module is used to summarize somatic mutation data, germline mutation analysis results, MSI status analysis results, copy number variation analysis results, TMB values ​​and tumor neoantigen prediction results to generate a panoramic mutation map of tumor whole exome sequencing.

[0125] The sequence alignment module receives raw whole-exome sequencing (WES) data from tumor samples and their paired normal samples. Its output is connected to the data preprocessing module. The data preprocessing module's output is connected to the inputs of the germline mutation analysis module, MSI status analysis module, copy number variation analysis module, and somatic mutation detection module. The outputs of these modules are connected to the results aggregation module. The somatic mutation detection module's output is connected to the inputs of the tumor neoantigen analysis module and the TMB analysis module. The outputs of these modules are also connected to the results aggregation module.

[0126] The system receives raw whole-exome sequencing (WES) data (FASTQ format) of tumor samples and their paired normal samples, as well as necessary sample metadata.

[0127] Starting Argo Workflows: After a user submits an analysis request, the system instantiates and starts a workflow instance on the Kubernetes cluster based on the preset Argo Workflows template and input parameters. Reference genomes, annotation databases, etc., are provided as shared resources to each analysis container.

[0128] Example:

[0129] A panoramic variation analysis was performed on a tumor sample from a hypothetical lung adenocarcinoma patient and its paired normal blood sample:

[0130] 1. Sample Information and Sequencing Data Preparation

[0131] Patient information: Virtual patient P001, diagnosed with lung adenocarcinoma.

[0132] Sample type:

[0133] Tumor sample (T): DNA extracted from lung adenocarcinoma tissue of patient P001.

[0134] Normal sample (N): DNA extracted from the peripheral blood of patient P001.

[0135] Sequencing type: Whole exome sequencing (WES), using the Illumina NovaSeq 6000 platform, PE150 sequencing mode.

[0136] Raw data: Generate FASTQ format file:

[0137] P001_T_R1.fastq.gz, P001_T_R2.fastq.gz

[0138] P001_N_R1.fastq.gz, P001_N_R2.fastq.gz

[0139] Reference data:

[0140] Reference genome: GRCh38 / hg38.

[0141] Gene annotation: ENSEMBL v112.

[0142] Known variant databases: dbSNP, ClinVar, COSMIC, gnomAD.

[0143] MSI loci: UCSC hg38 standard microsatellite loci set.

[0144] Target region: BED file corresponding to the whole exome sequencing (WES) capture kit.

[0145] 2. System Environment and Argo Workflows Configuration

[0146] Computing environment: Deployed on a local Kubernetes cluster (v1.21) containing 3 compute nodes (each node is configured with 32 CPU cores, 128GB RAM, and 1TB NVMe SSD).

[0147] Core software:

[0148] To use Docker (v20.10) as a container runtime, you need to install Docker on the Kubernetes node and configure the kubelet to correctly point to the Docker daemon.

[0149] Argo Workflows (v3.2) serves as the workflow engine.

[0150] Shared storage: Configure NFS as storage for Argo artifacts and input / output data.

[0151] Tool containerization: Figure 1All bioinformatics tools involved (BWA-MEM 0.7.17, Samtools 1.10, GATK 4.2.0.0 (including BaseRecalibrator, HaplotypeCaller, Mutect2), CNVkit 0.9.9, Manta 1.6.0, Strelka2 2.9.10, MSIsensor2 0.5, VarScan2 2.4.4, VCF2MAF (e.g., vcf2maf.pl v1.6.21), PVACTools (e.g., pVACseq module from v3.0.0) are packaged into independent Docker images and uploaded to a private Docker Registry. Each image contains the tool itself and all its dependencies.

[0152] Argo Workflows definition:

[0153] Create an Argo Workflows template file named tumor-wes-comprehensive-pipeline.yaml.

[0154] This YAML file strictly follows the flowchart and modular design described in the technical solution of this invention, defining each analysis step, including:

[0155] Input parameters: FASTQ file path, sample ID, reference genome path, annotation file path, etc.

[0156] Task dependencies: Clarify the order and parallel relationships of each step (for example, BaseRecalibrator can only be executed after BWA alignment; Mutect2, CNVkit, Manta, etc. can process and analyze ready BAM files in parallel).

[0157] Container image specification: Each step specifies the Docker image to use.

[0158] Commands and parameters: The specific commands and parameters executed within each container.

[0159] Input / output workpieces: Defines the data transfer between steps (such as BAM files, VCF files, MAF files, etc.).

[0160] 3. Implementation steps (automated via Argo Workflows)

[0161] (1) Workflow submission and initialization

[0162] Users submit `tumor-wes-comprehensive-pipeline.yaml` via the Argo CLI or UI, specifying the FASTQ file path for the P001 sample and other necessary parameters. The Argo Workflows engine instantiates the workflow on a Kubernetes cluster.

[0163] (2) Sequence alignment (Normal BWA-MEM, Tumor BWA-MEM)

[0164] Argo launches two BWA-MEM tasks in parallel to process the FASTQ files of P001_T and P001_N respectively, and aligns them to the hg38 reference genome.

[0165] After each BWA-MEM task is completed, the Samtools task is launched to perform sorting, format conversion (SAM to BAM), deduplication (MarkDuplicates, integrated into GATK or using Picard), and indexing, generating P001_T.sorted.dedup.bam and P001_N.sorted.dedup.bam.

[0166] (3) Data preprocessing (Normal BaseRecalibrator, Tumor BaseRecalibrator)

[0167] Argo launches two GATK BaseRecalibrator tasks in parallel to perform base quality score recalibration on P001_T.sorted.dedup.bam and P001_N.sorted.dedup.bam respectively, generating analysis-ready BAM files: P001_T.bqsr.bam and P001_N.bqsr.bam.

[0168] (4) Execute multiple mutation detection modules in parallel.

[0169] ① Germline mutation analysis:

[0170] GATK HaplotypeCaller performs germline variation detection on P001_N.bqsr.bam and outputs P001_N.germline.vcf.

[0171] VCF2MAF converts P001_N.germline.vcf to P001_N.germline.maf.

[0172] ②Somatic cell mutation detection (parallel):

[0173] Mutect2: Input P001_T.bqsr.bam and P001_N.bqsr.bam, output P001_somatic_mutect2.vcf.

[0174] VarScan2: Samtools mpileup generates a pileup file, and VarScan2 processes it to output P001_somatic_varscan.vcf.

[0175] Strelka2: Input P001_T.bqsr.bam and P001_N.bqsr.bam, output P001_somatic_strelka2.vcf.

[0176] Manta: Input P001_T.bqsr.bam and P001_N.bqsr.bam, output P001_somatic_manta_sv.vcf (containing SV information).

[0177] ③ Microsatellite instability analysis:

[0178] MSIsensor2 analyzes P001_T.bqsr.bam and P001_N.bqsr.bam, and outputs P001_msi_status.txt (containing MSI score and status).

[0179] ④CNV analysis:

[0180] CNVkit analyzes P001_T.bqsr.bam and P001_N.bqsr.bam (with normal samples as controls), and outputs P001_T.cns (copy number fragment file) and related charts.

[0181] (5) MAF Conversion (Parallel)

[0182] Convert P001_somatic_mutect2.vcf, P001_somatic_varscan.vcf, and P001_somatic_strelka2.vcf to MAF format: P001_somatic_mutect2.maf, P001_somatic_varscan.maf, and P001_somatic_strelka2.maf, respectively. This step will perform preliminary gene annotation and functional prediction (e.g., using VEP).

[0183] ⑹ TMB Analysis (Parallel)

[0184] Based on the MAF files generated in step (5), calculate the TMB value (such as the number of non-synonymous mutations per Mb coding region) and output P001_tmb_mutect2.txt, P001_tmb_varscan.txt, and P001_tmb_strelka2.txt.

[0185] (7) Neoantigen Analysis (PVACTools)

[0186] Assume that the HLA allele information of patient P001 has been obtained through HLA typing (e.g., HLA-A*02:01, HLA-B*07:02, etc.).

[0187] Select a high-quality somatic MAF file (e.g., P001_somatic_mutect2.maf) and HLA information as input to PVACTools (pVACseq) to predict potential neoantigen peptides and output P001_neoantigens.tsv.

[0188] (8) Results summary and report generation

[0189] Argo launches a result aggregation task, which contains a container of custom Python / R scripts.

[0190] Data collection: Collect all key result files generated upstream:

[0191] P001_N.germline.maf

[0192] P001_somatic_mutect2.maf,P001_somatic_varscan.maf,P001_somatic_strelka2.maf

[0193] P001_somatic_manta_sv.vcf

[0194] P001_T.cns and related chart information output by CNVkit

[0195] P001_msi_status.txt

[0196] P001_tmb_mutect2.txt

[0197] P001_neoantigens.tsv

[0198] 4. Expected Results and Beneficial Effects

[0199] Execution time: The entire process is expected to be completed within 6-10 hours (the exact time depends on the amount of data and the cluster load).

[0200] Output: A comprehensive, structured result file containing multiple variant types.

Claims

1. A panoramic variant mapping analysis method for tumor whole-exome sequencing based on Argo Workflows, characterized in that... Includes the following steps: S1. Using Argo Workflows, each tool used in the various analysis processes for variant type analysis of whole-exome sequencing raw data of tumor samples and their paired normal samples is packaged into an independent Docker container and defined as a task. S2. Use two parallel task branches to perform sequence alignment processing on the raw whole exome sequencing data. One task branch aligns the sequencing reads of the tumor sample to the reference genome to generate tumor sample sequence alignment data, and the other task branch aligns the sequencing reads of the normal sample paired with the tumor sample to the reference genome to generate normal sample sequence alignment data. S3. Use two parallel task branches to preprocess the data obtained from step S2, that is, the two task branches perform base quality score recorrection on the tumor sample sequence alignment data and the normal sample sequence alignment data, respectively. S4. Perform the following analysis and processing workflow in parallel on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data: (1) Perform germline mutation analysis on the sequence alignment data of normal samples after data preprocessing to obtain the germline mutation analysis results; (2) Copy number variation analysis was performed on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the copy number variation analysis results; (3) Perform MSI status analysis on the preprocessed tumor sample sequence alignment data and normal sample sequence alignment data to obtain the MSI status analysis results; (4) Perform somatic mutation detection on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data; S5. Based on the sequence alignment data of tumor samples after somatic mutation detection, the sequence alignment data of normal samples, and the patient's HLA typing information, predict tumor neoantigens and obtain tumor neoantigen prediction results. S6. Calculate the TMB value by comparing the sequence alignment data of tumor samples and normal samples after somatic mutation detection. S7. The results of somatic mutation analysis, germline mutation analysis, MSI status analysis, copy number variation analysis, TMB value and tumor neoantigen prediction are summarized to generate a panoramic mutation map of tumor whole exome sequencing.

2. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 1, characterized in that: In step S2, each task branch first uses the BWA-MEM tool to perform sequence alignment to generate a SAM file, and then uses the Samtools tool to convert the format of the SAM file, sort it, and index it to generate a BAM file.

3. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 2, characterized in that: In step S3, each task branch uses the BaseRecalibrator tool to perform base quality score recalibration and outputs a BAM file.

4. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 3, characterized in that: In step S4⑴, the Germline HaplotypeCaller tool is first used to perform germline SNV and Indel detection to generate a VCF file. Then, the VCF2MAF tool is used to convert the VCF file into an MAF file. Finally, the filtered germline mutation analysis results are input into the results summary.

5. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 4, characterized in that: In step S4⑵, the CNVkit tool is used to perform copy number variation analysis, and then the filtered copy number variation analysis results are summarized.

6. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 5, characterized in that: In step S4⑶, the MSI state analysis is performed using the MSIsensor2 tool.

7. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 6, characterized in that: Step S4(4) uses multiple tools to detect somatic SNVs / Indels and SVs in parallel. Specifically, the Mutect2 tool is used to detect somatic SNVs and Indels and generate a VCF file; the Samtools mpileup tool is used in parallel to generate a pileup format file, and then the VarScan2 tool is used to detect somatic SNVs and Indels and generate a VCF file; the Manta tool is used in parallel to detect SVs, and then the SVs are annotated, fusion gene analysis is performed, and the results are entered into the summary; at the same time, the Manta tool is used to detect SVs and generate candidate Indel intervals, which are then entered into the Strelka2 tool for somatic SNV and Indel detection and generate a VCF file. Simultaneously, the Strelka2 tool performs somatic SNV and Indel detection on the data after data preprocessing; finally, the VCF files output by the Mutect2, VarScan2, and Strelka2 tools are converted into MAF files in parallel.

8. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 7, characterized in that: In step S5, the tumor sample sequence alignment data and normal sample sequence alignment data after somatic mutation detection refer to the MAF files output and converted by the Mutect2, VarScan2 and Strelka2 tools, and the tumor neoantigens are predicted using the PVACTools tool.

9. The method for panoramic variant mapping analysis of tumor whole-exome sequencing based on Argo Workflows according to claim 8, characterized in that: In step S6, the sequence alignment data of the tumor sample and normal sample after somatic mutation detection specifically refers to the MAF file generated by converting the detection results output by the Mutect2 tool, the MAF file generated by converting the somatic SNV detection results output by the VarScan2 tool, and the MAF file generated by converting the somatic SNV detection results output by the Strelka2 tool. Based on the above MAF files, TMB calculation is performed in parallel, and the filtered somatic detection results are directly used to generate an HTML file summary of input results that users can directly view. The MAF files converted from the detection data output by the Mutect2 tool, the MAF files converted from the detection data output by the VarScan2 tool, and the MAF files converted from the detection data output by the Strelka2 tool are respectively input into the summary of results.

10. A system for analyzing the panoramic variant mapping of tumors based on Argo Workflows according to claim 1, characterized in that... include: The sequence alignment module is used to perform sequence alignment processing on the raw whole-exome sequencing data of tumor samples and their paired normal samples to obtain tumor sample sequence alignment data and normal sample sequence alignment data. The data preprocessing module is used to perform base quality score recorrection on tumor sample sequence alignment data and normal sample sequence alignment data; The germline mutation analysis module is used to perform germline mutation analysis on the preprocessed normal sample sequence alignment data. The copy number variation analysis module is used to perform copy number variation analysis on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data. The MSI status analysis module is used to perform MSI status analysis on preprocessed tumor sample sequence alignment data and normal sample sequence alignment data. The somatic mutation detection module is used to detect somatic mutations in preprocessed tumor sample sequence alignment data and normal sample sequence alignment data. The tumor neoantigen analysis module is used to predict tumor neoantigens based on sequence alignment data of tumor samples after somatic mutation detection, sequence alignment data of normal samples, and patient HLA typing information. The TMB analysis module is used to calculate TMB on the sequence alignment data of tumor samples and normal sample sequences that have undergone somatic mutation detection. The results summary module is used to summarize somatic mutation data, germline mutation analysis results, MSI status analysis results, copy number variation analysis results, TMB values ​​and tumor neoantigen prediction results to generate a panoramic mutation map of tumor whole exome sequencing. The input of the sequence alignment module is used to receive the raw whole-exome sequencing data of tumor samples and their paired normal samples. The output of the sequence alignment module is connected to the input of the data preprocessing module. The output of the data preprocessing module is connected to the input of the germline mutation analysis module, the MSI status analysis module, the copy number variation analysis module, and the somatic mutation detection module, respectively. The outputs of the germline mutation analysis module, the copy number variation analysis module, and the MSI status analysis module are connected to the result aggregation module. The output of the somatic mutation detection module is connected to the input of the tumor neoantigen analysis module and the TMB analysis module, respectively. The outputs of the tumor neoantigen analysis module and the TMB analysis module are connected to the result aggregation module, respectively.

Images