An HLA typing tool selection system, method, device and medium based on a typing tool calling agent

Abstract

This invention relates to the fields of bioinformatics and immunogenetics, and discloses an HLA typing tool selection system, method, device, and medium based on a typing tool invocation agent. The method includes: acquiring second-generation sequencing data and Sanger sequencing data from patients with autoimmune diseases; having a typing tool invocation agent invoke several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data, obtaining several HLA typing results; combining these with reference sequencing data to determine the performance evaluation results of each HLA typing tool; screening and optimizing HLA typing tools; and finally, obtaining the final HLA typing result. This invention, through data-driven tool optimization and gold standard validation, significantly improves the accuracy, reliability, and ability to discover new risk alleles in HLA typing in complex disease contexts, effectively reducing the cost of large-scale research, and has significant value for the study of the genetic mechanisms of autoimmune diseases and precision medicine.

Description

Technical Field

[0001] This invention relates to the fields of bioinformatics and immunogenetics, and in particular to an HLA typing tool selection system, method, device and medium based on typing tools calling intelligent agents. Background Technology

[0002] The human leukocyte antigen (HLA) gene, located on human chromosome 6, is the most polymorphic gene group and plays a central role in the adaptive immune response. The proteins it encodes are responsible for antigen processing and presentation. Numerous genetic studies have confirmed a close association between HLA alleles and susceptibility, disease progression, and clinical manifestations of various autoimmune diseases, such as systemic lupus erythematosus (SLE), rheumatoid arthritis, Sjögren's syndrome, and ankylosing spondylitis. Therefore, obtaining accurate, high-resolution HLA typing data is a prerequisite for in-depth analysis of the genetic mechanisms of these diseases, identification of risk biomarkers, and the realization of precision medicine.

[0003] Traditional HLA typing methods, such as serological typing, while highly accurate, typically suffer from limitations such as low throughput, high cost, cumbersome operation, and limited resolution (e.g., serological typing is binary), making them unsuitable for large-scale disease cohort studies and clinical screening. With the rapid development of next-generation sequencing (NGS) technology, the massive amounts of data generated through whole-exome sequencing, targeted sequencing, or RNA sequencing have provided new possibilities for high-throughput, high-resolution HLA typing. Correspondingly, various computational HLA typing tools have emerged, such as HISAT-genotype, OptiType, seq2HLA, HLA-HD, and HLA... LA, HLA-Ascan, etc. However, when these computational tools are applied to autoimmune disease research, the following significant limitations and challenges have been exposed:

[0004] First, there is a severe lack of consistency in genotyping results among different tools. Even when processing the same NGS data, different computational tools often produce HLA allele lists that differ significantly in both number and specific types. This high degree of inconsistency makes it difficult for researchers to determine which results are truly reliable, directly leading to uncertainty and non-reproducibility in the analysis results, and causing difficulties for downstream genetic association studies.

[0005] Secondly, existing tools have unstable and poor reproducibility in identifying known disease risk alleles. For example, the HLA-DRB1 risk allele in systemic lupus erythematosus... For example, at 15:01, when trying to validate using different tools, although most tools could indicate the direction of the risk effect, only a very small number of tools achieved statistical significance. This indicates that in the complex genetic context of autoimmune diseases, the performance of existing tools varies greatly, making it difficult not only to stably and reliably reproduce known genetic association signals, but also severely limiting their potential and reliability in discovering new risk genes or alleles.

[0006] Furthermore, there is a lack of systematic and authoritative performance evaluation systems for target disease populations. Currently, the vast majority of evaluation and comparative studies on HLA typing tools primarily use test datasets from healthy individuals or cancer patient cohorts. Because the distribution of HLA allele frequencies, linkage disequilibrium patterns, and gene region complexity are specific across different disease groups, performance conclusions drawn in a "general" or "other disease" context may not be applicable to autoimmune disease populations with particularly complex HLA genetic backgrounds.

[0007] In summary, existing NGS-based computer HLA typing methods face core shortcomings when applied to autoimmune disease research, such as inconsistent results, unstable performance, and lack of targeted evaluation criteria. There is an urgent need for a new method that can systematically solve the above problems and provide a reliable basis for tool selection for specific research scenarios. Summary of the Invention

[0008] This invention provides an HLA typing tool selection system, method, device, and medium based on typing tools calling intelligent agents, in order to overcome the shortcomings of existing technologies.

[0009] This invention provides an HLA typing tool selection system based on typing tool invoking an intelligent agent, comprising: The process coordination module is used to schedule the selection process of execution tools for various modules and intelligent agents within the system. The data acquisition module is used to acquire second-generation sequencing data of patients with specified diseases and reference sequencing data for calibrating or validating HLA typing results, based on sequencing conditions. The typing tool calls an intelligent agent to call several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data and obtain multiple HLA typing results. The performance evaluation module is used to determine the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in combination with reference sequencing data. The tool selection module is used to determine the preferred HLA typing tool based on the performance evaluation results.

[0010] The HLA genotyping tool selection system based on a genotyping tool invocation agent provided by the present invention further includes a tool performance database for storing performance evaluation records of each HLA genotyping tool under different sequencing conditions and preferred HLA genotyping tool tags. The performance evaluation records are generated by a performance evaluation module and are invoked by a tool selection module. The preferred HLA genotyping tool tags are used to indicate the HLA genotyping tool that performs optimally under specific sequencing conditions.

[0011] According to the present invention, an HLA typing tool selection system based on typing tool calling agent is provided, wherein the second-generation sequencing data includes any one or any combination of the following: whole exome sequencing (WES) data, MHC region targeted sequencing data, and RNA sequencing (RNA-seq) data; the reference sequencing data includes Sanger sequencing data.

[0012] According to the present invention, an HLA genotyping tool selection system based on a genotyping tool calling agent is provided, wherein the specified disease is an autoimmune disease, the reference sequencing data is Sanger sequencing data, and the Sanger sequencing data includes sequencing data for four-dimensional resolution genotyping of classic HLA class I loci and classic HLA class II loci (HLA-A, -B, -C, -DPB1, -DQA1, -DQB1, -DRB1) for patients with autoimmune diseases.

[0013] According to the present invention, an HLA typing tool selection system based on typing tool invocation agent is provided, wherein several HLA typing tools include any one or any combination of the following: HISAT-genotype, OptiType, seq2HLA, HLA-HD, HLA LA, HLA-Ascan.

[0014] It will be apparent to those skilled in the art that reference sequencing data used to calibrate or validate HLA genotyping results may include low-throughput genotyping methods for high-resolution validation, such as Sanger sequencing, sequence-specific primers, and sequence-specific oligonucleotide probes.

[0015] Because second-generation sequencing data has high throughput, it is prone to genotyping errors in highly polymorphic regions such as HLA. Sanger sequencing genotyping, on the other hand, has extremely high accuracy and is often regarded as the gold standard for four-dimensional resolution. Therefore, in this invention, Sanger sequencing is used to calibrate and evaluate the performance of four-dimensional resolution genotyping of HLA genotyping results based on second-generation sequencing data.

[0016] According to the present invention, an HLA typing tool selection system based on typing tool invocation agent is provided. The performance evaluation module is configured to evaluate the performance of the multiple HLA typing tools based on the multiple HLA typing results and in combination with reference sequencing results, using preset evaluation indicators to determine the performance evaluation results of each HLA typing tool. The preset evaluation indicators include any one of the following or any combination thereof: computational resource requirements, detection rate, accuracy, and robustness.

[0017] According to the present invention, an HLA typing tool selection system based on typing tool invocation agent is provided, wherein the expression for performance evaluation of each HLA typing tool is as follows: F （工具i,C） =w pos ×S pos(i,C) + w acc × S acc(i,C) + w robust ×S robust(i,C) + w res ×S res(i,C) , In the formula, w pos w acc w robust w res S represents the preset weight of each evaluation indicator. pos(i,C) S represents the detection rate score of HLA typing tool i under specific sequencing conditions C. acc(i,C) S represents the accuracy score of HLA genotyping tool i under specific sequencing conditions C. robust(i,C) S represents the robustness score of HLA genotyping tool i under specific sequencing conditions C. res(i,C) F represents the computational resource requirement score of HLA genotyping tool i under specific sequencing conditions C. （工具i,C） This represents the performance score of the HLA typing tool i under specific sequencing conditions C.

[0018] In the above expression, specific sequencing conditions refer to the sequencing conditions input by the user, including but not limited to: sequencing platform (e.g., Illumina, BGI / MGI, Ion Torrent, etc.; different platforms have different error patterns, which will affect the performance of HLA genotyping tools), sequencing data type (e.g., WES, WGS, MHC region-targeted sequencing, RNA-seq, etc.), sequencing depth, read length, library construction method, data quality indicators, target gene, and custom weights. In a specific implementation, if all sequencing data come from the same sequencing platform (e.g., Illumina), then the specific sequencing conditions include sequencing data type, read length, data depth, target gene, and custom weights.

[0019] The present invention provides an HLA genotyping tool selection system based on a genotyping tool invoking an agent. The system calculates a resource requirement score to reflect the efficiency of the HLA genotyping tool in terms of execution time and peak memory usage. The detection rate is defined as the percentage of samples for which the HLA genotyping tool successfully generates HLA allele type predictions for a given locus. Accuracy reflects the consistency between the HLA genotyping results and the gold standard (the proportion of correctly predicted alleles among successfully genotyped alleles at four-position resolution). Robustness is assessed based on different sequencing depths and read lengths.

[0020] In one embodiment, the system further includes a tool performance database for storing performance evaluation records and preferred HLA genotyping tool markers for each HLA genotyping tool under different sequencing conditions. The performance evaluation records are generated by the performance evaluation module and are used by the tool selection module. The performance evaluation records record the correspondence between different sequencing conditions and the performance scores of each HLA genotyping tool, including indicators such as genotyping accuracy, consistency, runtime, and resource consumption. The preferred HLA genotyping tool markers indicate the HLA genotyping tool that performs optimally under specific sequencing conditions. This tool performance database can be used to support the system of this invention in quickly selecting preferred tools in subsequent tasks, providing performance evaluation records for other follow-up work, and forming a long-term accumulated tool performance knowledge base.

[0021] In one implementation, the genotyping tool calling agent includes: an input processing unit for receiving and preprocessing second-generation sequencing data and tool operating parameters; a tool execution unit for calling multiple HLA genotyping tools and monitoring tool operating status; a result parsing unit for parsing the output files of each HLA genotyping tool and generating standardized HLA genotyping results; and a status feedback unit for feeding back tool operating status and error information to the process coordination module.

[0022] Specifically, the input processing unit is configured to: automatically identify the input data format, including FASTQ, BAM, or CRAM; perform format conversion according to the operational requirements of the target tool; and automatically generate the parameter configuration file required for tool operation. The tool execution unit is configured to invoke the HLA typing tool through at least one of the following methods: invoking a command-line program; invoking a containerized runtime environment; invoking the API interface provided by the tool; or invoking a local executable file. The result parsing unit is configured to parse the output formats of different HLA typing tools, including but not limited to: OptiType's result.tsv file; HLA-HD's HLAgenotype.txt file; HLA... The hla.json file for LA, etc.

[0023] The typing tool invoking agent provides a unified input interface, output interface, and status feedback interface, enabling the system to invoke different HLA typing tools in a consistent manner. After completing the typing task, the typing tool invoking agent sends the tool runtime, resource consumption, error information, and typing result quality indicators to the performance evaluation module for generating performance evaluation records and storing them in the tool performance database.

[0024] This invention also provides a method for selecting HLA typing tools based on typing tools calling intelligent agents, comprising: Receive sequencing conditions input by the user; According to the sequencing conditions, obtain second-generation sequencing data of patients with the specified disease and reference sequencing data for calibrating or validating HLA typing results; The typing tool calls an agent to call several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data, resulting in multiple HLA typing results; The performance evaluation module determines the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in conjunction with reference sequencing data. The tool selection module determines the preferred HLA typing tool based on the performance evaluation results.

[0025] This invention also provides a method for identifying susceptibility genes associated with autoimmune diseases. This method utilizes the HLA typing tool selection system based on the typing tool of this invention to invoke an intelligent agent, and includes: Acquire second-generation sequencing data of the target subject, wherein the target subject is a patient with an autoimmune disease; Analysis of sequencing conditions for second-generation sequencing data; The preferred HLA typing tool is determined by retrieving performance evaluation records and preferred HLA typing tool markers that match the sequencing conditions from the tool performance database. The performance evaluation records are generated by the HLA typing tool selection system based on the typing tool calling agent. The preferred HLA typing tool was used to perform HLA typing analysis on the second-generation sequencing data to obtain the target HLA typing results. Based on the target HLA typing results, an autoimmune disease association database is queried to identify susceptibility genes associated with the disease.

[0026] The present invention also provides an electronic device, including a processor and a memory storing a computer program, wherein the processor executes the computer program to implement the above-described HLA typing tool selection method based on typing tool invoking an agent, and / or the above-described method for identifying susceptibility genes associated with autoimmune diseases.

[0027] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described HLA typing tool selection method based on typing tool invoking an agent, and / or the above-described method for identifying susceptibility genes associated with autoimmune diseases.

[0028] The present invention also provides a computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is capable of executing the above-described HLA typing tool selection method based on typing tool invoking an agent, and / or the above-described method for identifying susceptibility genes associated with autoimmune diseases.

[0029] The present invention provides an HLA typing tool selection system, method, device, and medium based on typing tools calling intelligent agents, which can bring at least the following beneficial effects: This invention acquires NGS data from patients with a specified disease, then uses an intelligent agent to autonomously invoke multiple HLA typing tools. It introduces reference sequencing data (e.g., Sanger sequencing data) specific to the disease population for calibration or validation as the gold standard to evaluate the performance of HLA typing tools, thereby selecting the optimal HLA typing tool for that specific scenario. This fundamentally addresses the core pain points of existing tools, such as inconsistent performance in complex disease populations, lack of authoritative validation, and unreliable results, providing a high-quality, reliable HLA typing data foundation for subsequent genetic association analysis. For example, HLA genes in SLE exhibit extreme complexity, making HLA identification difficult for software. The optimal computer HLA typing tool for SLE research depends on the available NGS data types, the specific genes of interest, and computational resources.

[0030] This invention constructs a quantitative evaluation framework for HLA genotyping tools, integrating multiple dimensions such as computational resource requirements, detection rate, accuracy, and robustness. Based on real-world data, this framework objectively and quantitatively evaluates the performance of different HLA genotyping tools in specific NGS data types and autoimmune disease contexts, thereby selecting the optimal tool or tool combination best suited to the characteristics of the user's data. This overcomes the risk of unreliable analytical results due to inappropriate tool selection, providing users with a scientific basis for choice.

[0031] This invention is compatible with various NGS data sources, including whole-exome sequencing, whole-genome sequencing, MHC region-targeted sequencing, and RNA sequencing, and does not rely on a single sequencing platform. This flexibility ensures that the method can adapt to the actual data conditions of different research institutions or clinical projects, thereby flexibly determining the optimal genotyping scheme according to different research scenarios (such as data characteristics and resolution requirements), enhancing the universality and practical value of the method.

[0032] This invention, by introducing a gold-standard-based quantitative scoring framework, enables the stable and reliable screening of HLA typing tools that perform optimally in autoimmune disease cohorts. Application results demonstrate that only through this systematic evaluation and optimization process can new risk alleles (such as HLA-A for SLE) be successfully identified. (32:01), a finding that cannot be achieved with random selection tools. This provides a more powerful tool for a deeper understanding of the genetic mechanisms of autoimmune diseases and helps in discovering new genetic evidence.

[0033] This invention not only provides reliable technical support for a deeper understanding of the genetic basis of autoimmune diseases, and assists in genetic association research and biomarker identification; its highly reliable HLA typing data can also be directly used to construct disease risk prediction models, serving patient risk stratification, prognosis assessment and the formulation of individualized treatment strategies, thereby powerfully promoting the development of precision medicine in the field of autoimmune diseases. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0035] Figure 1 This is a flowchart illustrating an HLA typing tool selection method based on a typing tool calling an intelligent agent, as provided by the present invention.

[0036] Figure 2 This is one of the schematic diagrams comparing the performance of different HLA typing tools in processing different types of data in Example 1.

[0037] Figure 3 This is the second schematic diagram comparing the performance of different HLA typing tools in processing different types of data in Example 1.

[0038] Figure 4 This is the third illustration comparing the performance of different HLA typing tools in processing different types of data in Example 3. Detailed Implementation

[0039] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, embodiments of this invention, and should not be construed as limiting the invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention. In the description of this invention, it should be understood that the terminology used is for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0040] The execution subject of the HLA typing tool selection method based on typing tool invoking intelligent agent provided by the present invention can be any applicable terminal-side device or network-side device.

[0041] This invention provides a method for selecting HLA typing tools based on typing tools calling intelligent agents, which may include: S100: Receives sequencing conditions input by the user.

[0042] S110. Obtain second-generation sequencing data of patients with the specified disease and reference sequencing data for calibrating or validating HLA typing results according to the sequencing conditions.

[0043] In one embodiment, the second-generation sequencing data includes any one or any combination of the following: whole exome sequencing (WES) data, MHC region targeted sequencing data, and RNA sequencing (RNA-seq) data.

[0044] In one embodiment, the Sanger sequencing data includes sequencing data with four-position resolution genotyping of classic HLA class I loci and classic HLA class II loci (HLA-A, -B, -C, -DPB1, -DQA1, -DQB1, -DRB1) for patients with autoimmune diseases.

[0045] S120. The genotyping tool calls an agent to call several HLA genotyping tools to perform HLA genotyping analysis on the second-generation sequencing data, and obtain multiple HLA genotyping results.

[0046] In one embodiment, several HLA typing tools include any one or any combination of the following: HISAT-genotype, OptiType, seq2HLA, HLA-HD, HLA LA, HLA-Ascan.

[0047] S130. The performance evaluation module determines the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in conjunction with reference sequencing data.

[0048] In one embodiment, S130 may include: Based on several HLA typing results and combined with Sanger sequencing data, the performance of several HLA typing tools was evaluated using preset evaluation indicators to obtain the performance evaluation results of each HLA typing tool. The preset evaluation indicators include any one of the following or any combination thereof: computational resource requirements, detection rate, accuracy, and robustness.

[0049] In one embodiment, the expression for performance evaluation is: F （工具i,C） =w pos ×S pos(i,C) + w acc × S acc(i,C) + w robust ×S robust(i,C) + w res ×S res(i,C) , In the formula, w pos w acc w robust w res S represents the preset weight of each evaluation indicator. pos(i,C) S represents the detection rate score of HLA typing tool i under specific sequencing conditions C. acc(i,C) S represents the accuracy score of HLA genotyping tool i under specific sequencing conditions C. robust(i,C) S represents the robustness score of HLA genotyping tool i under specific sequencing conditions C. res(i,C) F represents the computational resource requirement score of HLA genotyping tool i under specific sequencing conditions C. （工具i,C） This represents the performance evaluation score of the HLA genotyping tool i under specific sequencing conditions C.

[0050] S140. The tool selection module determines the preferred HLA typing tool based on the performance evaluation results.

[0051] Optionally, the method may further include step S150, which includes sending the performance evaluation results generated in S140 and the preferred HLA typing tools generated in S140 to the tool performance database, and writing them into the tool performance database as performance evaluation records of each HLA typing tool under different sequencing conditions and preferred HLA typing tool tags.

[0052] The present invention will be further described below through specific embodiments.

[0053] Example 1: Selecting HLA typing tools based on sequencing conditions See Figure 1The method flowchart illustrates how this embodiment performed high-coverage whole-exome sequencing (WES), MHC-targeted sequencing, and RNA sequencing (RNA-seq) on an SLE cohort of 123 patients, comprehensively evaluating the performance of six widely used public HLA typing tools: HISAT-genotype, OptiType, seq2HLA, HLA-HD, and HLA. HLA, HLA-Ascan, and Sanger sequencing genotyping results were used as the gold standard. The performance of each HLA genotyping tool was evaluated across multiple dimensions, including computational resource usage, detection rate, accuracy, and robustness under different sequencing depths, read lengths, and alignment algorithms. Furthermore, the performance of various combinations of different read alignment tools and HLA genotyping algorithms was systematically evaluated. In addition, a quantitative scoring framework was used, integrating multiple performance metrics into a flexible weighted model, aiming to provide objective, data-driven guidance for selecting the most suitable HLA genotyping tool for different data types.

[0054] (1) Sample collection and data generation SLE patients are recruited from the CSTAR registry. Inclusion criteria include meeting the 2012 Systemic Lupus International Collaborative Clinic (SLICC) classification criteria or the 2019 American College of Rheumatology / European League Against Rheumatism (ACR / EULAR) criteria.

[0055] Blood samples were collected from participants. For WES, exome sequences were enriched from 0.4 μg of genomic DNA extracted from peripheral blood using the Agilent Liquid Capture System (AgilentSureSelect Human All Exon V6), following the manufacturer's protocol. The captured library was enriched by polymerase chain reaction (PCR) and indexed for sequencing. Products were purified using the AMPure XP system (Beckman Coulter) and quantified using the Agilent High Sensitivity DNA Assay on an Agilent Bioanalyzer 2100 system. The DNA library was sequenced at 150 bp paired ends using Illumina NovaSeq. For MHC-targeted sequencing, capture probes covering a total of 4.97 Mb (including a 3.37 Mb core MHC region and a 1.6 Mb extended region) were used to achieve deep, near-complete coverage of the MHC region. 3 mg of genomic DNA was fragmented and hybridized with the capture probes according to the manufacturer's protocol (Roche NimbleGen). The resulting libraries were sequenced on an Illumina HiSeq 2500 or HiSeq 4000 platform, generating 100 bp paired-end reads. For RNA-seq sequencing, red blood cells were lysed using potassium ammonium chloride lysis buffer (Gibco). Total RNA was extracted from peripheral blood samples using the PAXgene® Blood RNA Kit (PreAnalytiX). VAH™ mRNA Capture Beads (YEASEN) were used to isolate mRNA from the total RNA. Normal library construction was performed when the mRNA amount was in the range of 10 to 40 ng. Library construction was performed using the Hieff NGS Ultima Dual-mode mRNA Library Prep Kit for Illumina (YEASEN). The products were amplified by PCR using a Thermal cycler S1000 (Bio-Rad) to generate the library. Finally, a strand-specific library with rRNA removed was constructed and then quality controlled (QC) was performed by qPCR (StepOne Plus (ABI)) to ensure a concentration greater than 3 nM. Subsequent sequencing was performed using QC-compliant library arrangement. Sequencing was performed on a Novaseq 6000 (Illumina) in 150 bp paired-end mode.

[0056] All sequencing data were output in FastQ format and quality controlled using FastQC (v 0.11.9) with default parameters; all samples passed QC. Four-position HLA typing was performed using Sanger sequencing-based genotyping (SBT), and the resulting genotypes were established as the gold standard reference for subsequent accuracy assessments.

[0057] (2) HLA typing tools and their implementation This embodiment evaluates six widely used publicly available computational tools for HLA typing from NGS data: HISAT-genotype (v 1.3.2), OptiType (v 1.3.1), seq2HLA (v 2.3), HLA-HD (v1.6.1), and HLA... HISAT-genotype (v 1.0.3) and HLAscan (v 2.1.4). Each tool uses a different version of the built-in IPD-IMGT / HLA reference database and is capable of genotyping various genes (Table 1). Each tool was installed and executed according to the developer's documentation, typically using default parameters. The tools were run using WES, MHC targeted sequencing, and RNA-seq data as appropriate input formats (FASTQ or BAM), according to the requirements of each tool (Table 1). [Although HISAT-genotype and HLAscan claim to support multiple input formats, only one format is actually valid: FASTQ for HISAT-genotype and BAM for HLAscan]. For tools requiring pre-aligned input (HLAscan and HLA...), ... The reads were aligned to the human reference genome version GRCh38 / hg38 using Burrows-Wheeler Aligner software (v 0.7.17). The bam files were sorted using Samtools (v 1.16.1). All recognizable genes were detected by each software, but SBT-based genotyping was performed only for the classic class I (HLA-A, -B, -C) and class II (HLA-DPB1, -DQA1, -DQB1, -DRB1) loci.

[0058] Table 1

[0059] (3) Performance evaluation indicators The performance of each HLA genotyping tool was evaluated based on several metrics, including computational resource requirements, detection rate, accuracy, and robustness across different sequencing depths, read lengths, and alignment tools. Computational resource requirements were assessed by recording the execution time and peak memory usage of each tool in a controlled computational environment. Positive rate, or detection rate, was defined as the percentage of samples for which the tool successfully generated HLA allele type predictions for a given locus. Consistency with the gold standard (the proportion of correctly predicted alleles among successfully genotyped alleles at four-position resolution) was the primary metric for measuring accuracy. To evaluate the robustness of the HLA genotyping tools to variations in sequencing data characteristics, simulated datasets were created from the original FASTQ files. Reads were randomly downsampled using seqtk (v 1.4) to systematically reduce sequencing depth, generating datasets corresponding to 1, 5, 10, 25, 50, and 75-fold coverage. Similarly, reads were computationally pruned to shorter lengths (50, 75, 100, and 125 bp) using seqtk to investigate the impact of read length. Each evaluated HLA genotyping tool was then applied to these modified datasets, and the resulting HLA genotyping accuracy was calculated. To determine compatibility with different alignment tools, sequencing data were aligned to the same reference genome using BWA, bowtie2, HISAT2, and STAR. HLAAscan and HLA were used. LA performs HLA typing on the outputs of each comparator and compares the typing accuracy.

[0060] (4) Quantitative scoring framework To provide researchers with an objective, data-driven approach to select the most suitable HLA genotyping tool based on their specific experimental context, this embodiment develops a quantitative scoring framework. This framework integrates key performance indicators from the benchmark analysis in this embodiment into a flexible weighted model that generates customized recommended scores for each tool based on user-defined parameters, including sequencing platform, HLA gene of interest, sequencing depth, and read length.

[0061] Under specific sequencing conditions (C), the overall performance score of a specific genotyping tool (i) is defined by the following formula: F （工具i,C） =w pos ×S pos(i,C) + w acc × S acc(i,C) + w robust ×S robust(i,C) + w res ×S res(i,C) , In the formula, w pos w acc w robust wres S represents the preset weight of each evaluation indicator. pos(i,C) S represents the detection rate score of HLA typing tool i under specific sequencing conditions C. acc(i,C) S represents the accuracy score of HLA genotyping tool i under specific sequencing conditions C. robust(i,C) S represents the robustness score of HLA genotyping tool i under specific sequencing conditions C. res(i,C) The S represents the computational resource requirement score for HLA genotyping tool i under specific sequencing conditions C, where each S is a standardized score. While this embodiment provides a default set of weights based on common research priorities, the framework is designed to be flexible, allowing users to customize these weights to better reflect their specific analytical objectives. The detection rate (Spos) and accuracy (Sacc) scores in this embodiment are obtained by minimizing and maximizing their raw values ​​to a range of 0 to 1.

[0062] Robustness score S robust The stability of quantification tool performance to changes in sequencing depth and read length. It is the depth fraction S. depth and length fraction S length A weighted combination after min-max normalization is performed. To calculate these user-specified sub-scores falling between the test data points in this embodiment, linear interpolation is used. For example, performance at a given sequencing depth is estimated based on the performance of the two closest test depths. For depths or lengths outside the test range of this embodiment, the value of the closest test endpoint is used. To separate the impact of these variations from the tool's baseline performance, the interpolation (calculated as the product of positivity rate and accuracy) is calculated relative to its performance on the original dataset before min-max normalization.

[0063] Calculate resource score S res The evaluation tool assesses its efficiency in terms of execution time and memory usage. It is a standardized time score S. time and memory score S mem The weighted average. Since lower execution time (T) and peak memory usage (M) are preferred, these values ​​are inverted after min-max normalization to conform to the "higher is better" convention for other metrics. The time score is calculated as S. time (i) = 1 - (T i - T min ) / (T max - T min The calculation method for memory score is similar.

[0064] As a first step, the framework filters tools based on their compatibility with the target gene specified by the user. For the remaining tools, this embodiment applies a set of default weights, prioritizing accuracy and robustness in genetic association studies: w acc =0.70, w pos =0.20, w robust =0.10 and w res =0.

[0065] (1) Overview of benchmark testing and tool characteristics This embodiment uses sequencing data from 123 SLE patients to benchmark six widely used public computational tools, with sample sizes of 115 for WES, 73 for MHC targeted sequencing, and 56 for RNA-seq. The genotyping tools evaluated required different input formats (FASTQ or BAM), with four supporting RNA-seq data. All genotyping tools performed four-position HLA typing, while HISAT-genotype and OptiType achieved higher resolutions of 8 and 6 positions, respectively. Each genotyping tool used a built-in alignment tool and different versions of the IPD-IMGT / HLA reference database. The number of HLA genotypes identified by each tool varied considerably, ranging from 3 to 29 across different platforms and genotyping tools. Most programs were able to detect classic HLA genes such as A, B, C, DPA1, DPB1, DQA1, DQB1, DRA, and DRB1. In contrast, non-classical genes, including class I genes (E, F, G, H, J, K, L, P, V) and class II genes (DMA, DMB, DOA, DOB, MICA, MICB, TAP1, TAP2), were detected by only three tools. Notably, HLA-DRB2-9, associated with autoimmune diseases, was fully recognized only by HLA-HD (Table 1).

[0066] First, the basic operational performance of the software was evaluated. Significant differences in the computational resources required by different fracturing tools were observed (Figures 2A-C). Specifically, HLA... HLA and HISAT-genotype consistently demanded the highest processing time per sample (CPU hours) across all evaluated data types. Peak memory usage also varied considerably. LA and HISAT-genotype again exhibited significantly higher peak memory usage than other tools (Figure 2D-F). Regarding data types, in the dataset of this example, the average processing time for all tools was lowest in WES (1.27 cpu hours), followed by MHC targeted sequencing (1.67 cpu hours), while peak memory usage was lowest in WES (6.59 GB), followed by RNA-seq (6.83 GB). Next, this example analyzed classic class I and II genes, and the allelic genotyping consistency generated by each tool revealed different patterns (Figure 2G-L). Among all three data types, class I alleles showed higher average consistency than class II (WES 65.80% vs. 54.39%, MHC 75.35% vs. 60.24%, RNA-seq 92.62% vs. 68.17%), although the number of tools available for assessing class I was greater than that for class II, making complete consistency less likely. For class I and II alleles, RNA-seq data showed the highest average consistency, followed by WES, and then MHC targeted sequencing. This pattern may be attributed to the limited availability of tools for evaluating RNA-seq input. For HLA class I genes, OptiType showed the highest consistency across all data types, achieving (79.75%), (88.89%), and (100%) consistency in MHC, WES, and RNA-seq data, respectively, followed by HISAT-genotype with (68.48%), (83.72%), and (94.03%) consistency, respectively. For class II genes, HLA... LA showed the highest consistency, with 62.16% and 72.46% consistency in MHC and WES data, respectively, followed by HLA-HD, with 56.10%, 64.94%, and 73.85% consistency in MHC, WES, and RNA-seq data, respectively. In contrast, seq2HLA frequently identified alleles inconsistent with other tools. This unique calling pattern was most evident when using WES data, with nearly half of its predictions (47.06% for class I and 52.42% for class II) being unique.

[0067] (2) HLA typing performance indicators To evaluate the genotyping accuracy of each software, this example uses SBT sequencing, the gold standard, to validate class I and class II classic genes in all samples. The positive detection rate, defined as the percentage of samples for a given locus where the tool successfully generates an HLA allele type prediction, is typically high for most tools when analyzing WES and RNA-seq data. Figure 3BC). These rates typically exceed those observed when using MHC-targeted sequencing data ( Figure 3 A). Accuracy, defined as the proportion of correctly predicted alleles among successfully genotyped alleles at four-dimensional resolution, also varies depending on the tool and data type. Figure 3 DF). The average genotyping accuracy was highest in RNA-seq data (91.50%), followed by whole exome sequencing (87.24%), and lowest in MHC targeted sequencing (81.86%). Figure 3 (G). When specifically evaluating the performance of WES data for class I and II loci, different leaders emerged. For class I genes, HISAT-genotype demonstrated the highest accuracy using either WES (96.27%) or RNA-seq data (99.11%), with OptiType ranking second (96.08% for WES and 97.62% for RNA-seq). For MHC targeting data, OptiType ranked first (95.20%), followed by HISAT-genotype (89.60%). For class II genes, HLA-HD achieved the highest accuracy using either WES (93.72%) or RNA-seq data (99.55%), followed by HISAT-genotype (92.11% for WES and 89.51% for RNA-seq). For MHC targeting data, HISAT-genotype's accuracy far surpassed all other tools (92.93%). Taking both gene types into account, HISAT-genotype provides the highest overall genotyping accuracy when using WES data as input, while HLA-HD achieves the highest accuracy when using MHC-targeting or RNA-seq data.

[0068] (3) Robustness of sequencing parameters and alignment software To evaluate the robustness of different tools under various sequencing conditions, this example assesses their performance at a range of sequencing depths and read lengths. The results show that performance is generally robust when sequencing depth exceeds 25-fold, or RNA-seq data exceeds 5 Gb, with accuracy tending to stabilize or slightly increase at greater depths. Figure 3 JL). When using MHC targeting data, the genotyping accuracy of HISAT-genotype, HLA-HD, and HLA-Ascan plateaued at a sequencing depth of 50-fold, with no substantial improvement observed beyond that depth. In contrast, HLA... LA and OptiType saturated at lower depths of 10x and 25x, respectively. Notably, the accuracy of seq2HLA did not plateau with increasing depth, which may be attributed to its lower baseline accuracy. Figure 3 J). For WES data, all tools except HLAscan and seq2HLA are stable at 25x. HLAscan saturates at 50x, while seq2HLA does not show saturation even at the highest depth, although it achieves relatively high accuracy at 100x. Figure 3 K). In RNA-seq data, HLA-HD achieved high accuracy with only 1 Gb of data. The accuracy of HISAT-genotype and seq2HLA plateaued at 5 Gb, while OptiType showed a significant improvement at 10 Gb. Figure 3 L). To assess the impact of read length, the sequencing data were systematically pruned into a series of five datasets, starting at 50 bp and increasing by 25 bp (L). Figure 3 The results showed that, within the tested range, overall accuracy was minimally dependent on read length for most evaluation tools. Regardless of data type, HLA-HD and HLA-Ascan could not handle sequencing data with read lengths less than 100 bp. For MHC-targeted data, OptiType's accuracy increased when read lengths reached 100 bp, while other tools showed a more stable pattern (MO). Figure 3 M). On the other hand, among all three data types tested, the accuracy of seq2HLA may decrease with increasing read length. Figure 3 Similarly, when using WES data, HLAScane's genotyping accuracy decreases at a length of 125 bp, while the accuracy at 150 bp is similar to that at 100 bp. This is likely due to transitional artifacts caused by increased alignment ambiguity at a length of 125 bp.

[0069] Furthermore, since software supporting BAM input does not require a specific alignment unit, this embodiment compares the genotyping accuracy of various alignment and genotyping procedures (Table 1). For MHC targeted sequencing data, BWA and HLA... The LA combination achieved the highest accuracy (82.85%), followed by the BWA and HLA-Ascan combination (82.36%). Figure 3N). Although the differences between the classification tools were small, BWA consistently outperformed other comparators. For WES data, the combination of Bowtie2 and HLAsccan achieved the highest accuracy (91.82%). Figure 3 In this case, HLA-Ascan is significantly superior to HLA. LA, and Bowtie2 and BWA both outperformed HISAT2 and STAR. A notable exception was STAR compared to HLA. When LA pairs, it constitutes all HLA-based... The most efficient process in the combination of LA indicates enhanced compatibility between the alignment unit and specific genotyping algorithms. Tool-aligner compatibility also varies; Bowtie2 is not compatible with HLA. LA, but works well with HLAScane.

[0070] Based on the above results, the framework can be made into a web-based interface (such as a web page) for application in large-scale clinical cohorts of real-world patients. Furthermore, various HLA typing tools can be encapsulated into a unified typing tool invocation agent, enabling the system to perform multi-tool typing tasks in a modular, automated, and scalable manner. This significantly improves the system's robustness, maintainability, and large-scale application capabilities, providing crucial technical support for subsequent performance evaluation, optimal tool selection, and disease susceptibility gene identification.

[0071] Example 2: Construction of the Fractalization Tool's Invocation of Intelligent Agents and Tool Performance Evaluation Database (a) Encapsulation of the intelligent agent called by the fractal tool In this invention, an agent refers to a software entity that encapsulates specific processing logic and possesses basic capabilities for input perception, internal processing, and result output.

[0072] Building upon Example 1, this example encapsulates multiple HLA typing tools into a typing tool invocation agent for unified invocation of these tools within the system, and allows for scheduling by the process coordination module. The typing tool invocation agent is implemented in software, with its core module developed using Python and containerized to ensure consistency and portability of the runtime environment. The agent adopts a modular and plug-in architecture, enabling scalable invocation and system-level scheduling of multiple HLA typing tools through a unified interface. The encapsulation process of this agent is as follows: Step 1: Encapsulate the input processing unit During the encapsulation process, an input processing unit is first constructed to receive and preprocess second-generation sequencing data and tool execution parameters. This unit performs the following functions during encapsulation: integrates data format recognition logic for automatically identifying different formats such as FASTQ, BAM, and CRAM; integrates a format conversion module to perform necessary format conversions according to tool requirements; integrates a parameter parsing module to parse execution parameters such as read length, sequencing platform, and target gene; designs a unified input path and file structure specification to ensure that data from different sources can be input in a standardized manner; and encapsulates the preprocessed data into standardized input objects and provides an interface for passing them to the tool execution unit.

[0073] Step 2: Packaging Tool Execution Unit During the encapsulation process, a tool execution unit is built, enabling it to call various HLA typing tools and monitor their runtime status. This includes OptiType, HLA-HD, and HLA... Tools such as LA, HISAT-genotype, and seq2HLA are encapsulated as callable plugins; a runtime status monitoring module is integrated to monitor runtime, exit codes, resource usage, etc.; an exception handling mechanism is integrated to automatically retry or switch running modes when the tool fails; and a raw fractal result file is generated after the tool finishes running and then passed to the result parsing unit.

[0074] Step 3: Encapsulate the result parsing unit During the encapsulation process, a result parsing unit is constructed to parse the output files of various HLA genotyping tools and generate standardized HLA genotyping results. The encapsulation includes: an integrated output file format recognition module to identify the output structure of different tools; integrated parsing logic to extract HLA-A, HLA-B, HLA-C, HLA-DRB1, and HLA-DQB1 allele results; an integrated confidence score extraction module to read confidence scores or rating information from the tool output; an integrated runtime log and performance indicator extraction module; and the encapsulation of the parsed results into a unified, standardized HLA genotyping result object, which is then returned to the process coordination module.

[0075] Step 4: Encapsulate the status feedback unit During the encapsulation process, a status feedback unit is constructed to provide feedback on the tool's running status and error information to the process coordination module. The encapsulation includes: defining a unified running status feedback interface (success, failure, warning, etc.); integrating an error type identification module to classify exceptions that occur during tool operation; integrating an error logging module to record error stacks, runtime environment information, etc.; integrating a metadata collection module to collect performance metrics such as runtime and CPU / memory consumption; and returning the above status information to the process coordination module in a standardized format for scheduling optimization and exception handling.

[0076] Step 5: Register the agent After encapsulation, the agent that invokes the classification tool is registered in the system's agent directory, enabling it to be discovered and scheduled by the process coordination module, supporting future expansion with new tool plugins or tool version updates. Once registered, the agent can participate as a core component of the system in multi-tool classification, performance evaluation, and tool selection processes.

[0077] By leveraging intelligent agents, especially in large-scale clinical cohort applications in the real world, runtime can be significantly reduced, typing success rates can be improved, consistency between tools can be enhanced, and batch, pipelined processing can be supported.

[0078] (II) Construction of the Tool Performance Database To facilitate data-driven decision-making in various scenarios, a tool performance database can be built to store performance evaluation records of various genotyping tools under different sequencing conditions, as well as the labeling of preferred genotyping tools. This will provide performance knowledge support for subsequent analysis systems and indicate the tools that perform best under specific sequencing conditions.

[0079] 1. Data Sources: The database is constructed based on the following two types of data: (1) Second-generation sequencing data (NGS), including but not limited to: whole exome sequencing (WES), whole genome sequencing data, MHC region targeted sequencing, and RNA-seq. These data come from patients with specified diseases (such as autoimmune diseases). (2) Reference sequencing data: used to calibrate or validate HLA typing results, including four-resolution Sanger sequencing data or other high-accuracy sequencing methods.

[0080] 2. Construction Process: The construction of the tool performance database is uniformly scheduled by the system's process coordination module, and specifically includes the following steps: Step 1: Sequencing Condition Resolution: The system receives user-inputted or preset sequencing conditions, including, for example, sequencing platform, sequencing data type, data read length, data depth, target gene, and custom weights. The system uses these conditions as key fields in the database index.

[0081] Step 2: Perform multi-tool genotyping: Based on the sequencing conditions, call multiple HLA genotyping tools to genotype the same batch of second-generation sequencing data and output multiple HLA genotyping results.

[0082] Step 3: The performance evaluation module evaluates the tool's performance based on multiple HLA genotyping results and reference sequencing data. Evaluation metrics include: accuracy, detection rate, robustness, and computational resource requirements. The system generates a comprehensive performance score for each tool.

[0083] Step 4: Generate Performance Evaluation Records: The system writes the following information into the performance evaluation records: sequencing conditions (sequencing data type, read length, data depth, target gene, and custom weights, etc.), tool name, various performance indicators, overall score, calibration results of reference sequencing data, genotyping success rate, runtime, and other metadata. Each record corresponds to "performance of a specific tool under specific sequencing conditions".

[0084] Step 5: Write to the tool performance database: The system writes the performance evaluation records and the preferred classification tool tags to the tool performance database.

[0085] 3. Dynamic database updates: The system automatically schedules update tasks through the process coordination module. After each update, the system will recalculate the comprehensive score and automatically update the preferred classification tool flags to ensure that the database always keeps the latest preferred tool information.

[0086] Example 3: Application of the HLA typing tool selection system of the present invention in a large-scale real-world SLE patient cohort. To evaluate the consistency and real-world applicability of the HLA typing tool selection system of the present invention, this embodiment shifts from a previously controlled benchmark test in a small cohort to a large-scale clinical cohort containing 1,250 SLE patients with WES data.

[0087] First, six HLA typing tools were used to genotype the classic HLA genes (A, B, C, DPB1, DQA1, DQB1, DRB1) in 1,250 SLE samples. Each tool was run at its default parameters.

[0088] To evaluate the performance of each tool in association analysis, the cohort was stratified based on the presence or absence of lupus nephritis (LN) comorbidities (347 individuals in the LN group and 903 individuals in the non-LN group). For each HLA gene, each multi-allelic locus was decomposed into a series of biallelic loci, and the association between each common allele (alleles with a minor allele frequency greater than 1%) and LN susceptibility was tested separately. Association analysis was performed using PLINK (v1.9) software. In PLINK, additive logistic regression analysis was used, with sex and age adjusted as covariates. The performance of each tool was compared to see if it could reproduce the widely validated LN risk alleles (HLA-DRB1 was used in this example). 15:01) The association with LN risk, and its typing performance.

[0089] Based on historical data from the tool performance database, the HISAT-genotype tool (e.g., v1.3.2; https: / / daehwankimlab.github.io / hisat-genotype / ) was selected for further analysis of the 1,250 SLE patient cohort. Through the aforementioned association analysis, the system screened for HLA alleles potentially associated with LN risk. A significance threshold was set at P < 0.05. The analysis in this embodiment revealed significant differences in the output results of the six tools. Most notably, this embodiment observed extensive variability in the number of HLA alleles invoked by each tool across the entire cohort (…). Figure 4 A). Except for HLA-DQA1, Seq2HLA identified the most unique alleles at each locus, while OptiType reported the fewest class I alleles. When the cohorts were stratified by LN status, most tools invoked more unique alleles in the larger non-LN group, a finding likely attributable to their larger sample size (LN (n=347) vs. non-LN (n=903)). Figure 4 B). More importantly, when tested, they replicated HLA-DRB1. When considering the ability of these tools to correlate typically with LN risk at 15:01, their performance showed a significant divergence. Figure 4 C). Although all five applicable tools indicated a risk effect (odds ratio > 1), only HISAT-genotype and Seq2HLA produced statistically significant results (p < 0.05).

[0090] To guide the selection of the optimal HLA genotyping tool for downstream analysis within this cohort, this embodiment provides a quantitative scoring framework tailored to the characteristics of this dataset, designed to identify the most suitable tool for accurate and efficient genotyping. The parameters for this scenario are defined as: platform = WES, average sequencing depth = 183 (times), average read length = 150 bp. This embodiment calculates a composite score (if applicable) for all seven target HLA loci (HLA-A, -B, -C, -DPB1, -DQA1, -DQB1, and -DRB1) for each of the six evaluation tools. Using default weights, emphasizing accuracy (w=0.70) and detection rate (w=0.20) in genetic association studies, an overall performance score is calculated for each tool based on their performance in benchmark analyses. The performance scores for each component, including detection rate, accuracy, robustness, and computational resources, as well as the final composite score, are summarized in Supplementary Table 2. HISAT-genotype achieved the highest overall score of 1, demonstrating an excellent balance of high accuracy, robust performance on variable depth data, and reasonable computational efficiency for this dataset. In contrast, tools such as seq2HLA scored lower, primarily due to their lower accuracy. Platforms developed using this framework (e.g., web-based interfaces) allow users to input key parameters of their datasets, including sequencing platform, depth, read length, and data type, and generate a software-recommended ranking list, providing a customized and actionable starting point for HLA genotyping.

[0091] Table 2

[0092] Based on the preferred genotype markers in the tool performance database, this embodiment selected HISAT-genotype to perform the final HLA typing in a cohort of 1,250 SLE patients. Subsequent association analysis identified HLA-A. 32:01 is a putative risk allele for LN, and its observed frequency in the LN patient group (2.71%) was significantly higher than that in the non-LN group (1.51%), corresponding to a substantial increase in disease risk (OR [95% CI] = 2.05 [1.09–3.86], P = 0.025) (see [link to relevant documentation]). Figure 4 (D) This finding underscores the value of the systematic benchmarking and scoring framework of this invention in enabling robust downstream biological discoveries.

[0093] In summary, HISAT-genotype achieved the highest overall accuracy using both WES (93.85%) and MHC-targeted data (91.59%), while HLA-HD demonstrated superior performance using RNA-seq data (98.34%). For specific genotyping categories, OptiType and HLA-HD showed the highest accuracy for class I and class II genotyping, respectively. Overall, these tools generally maintained robust performance across different read lengths, and accuracy improved further when sequencing depth exceeded 25X. Alignment compatibility assessments indicated that Bowtie2 and BWA performed superiorly when used with HLA-Scan, while STAR performed better with HLA-Scan. LA matching yielded the best results. When WES data were applied to an independent SLE cohort comprising 1,250 patients, these tools exhibited considerable variability in allele detection. Guided by the quantitative scoring framework of this embodiment, HISAT-genotype was selected as the best-performing tool in this cohort and identified HLA-A. 32:01 is a novel putative risk allele for lupus nephritis (LN).

[0094] The HLA typing tool selection system based on typing tool invoking intelligent agent provided by the present invention is described below. The HLA typing tool selection system based on typing tool invoking intelligent agent described below can be referred to in correspondence with the HLA typing tool selection method based on typing tool invoking intelligent agent described above.

[0095] The HLA typing tool selection system based on typing tool invoking intelligent agent provided by this invention may include: The process coordination module is used to schedule the selection process of execution tools for various modules and intelligent agents within the system. The data acquisition module is used to acquire second-generation sequencing data of patients with specified diseases and reference sequencing data for calibrating or validating HLA typing results, based on sequencing conditions. The typing tool calls an intelligent agent to call several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data and obtain multiple HLA typing results. The performance evaluation module is used to determine the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in combination with reference sequencing data. The tool selection module is used to determine the preferred HLA typing tool based on the performance evaluation results.

[0096] The electronic device provided by this invention may include: a processor 810, a communications interface 820, a memory 830, and a communication bus 840, wherein the processor 810, the communications interface 820, and the memory 830 communicate with each other through the communication bus 840. The processor 810 can call logical instructions in the memory 830 to execute the steps of the above-described method for selecting an HLA typing tool based on a typing tool and / or the above-described method for identifying susceptibility genes related to autoimmune diseases.

[0097] Furthermore, the logical instructions in the aforementioned memory 830 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0098] On the other hand, the present invention also provides a computer program product, the computer program product comprising a computer program that can be stored on a non-transitory computer-readable storage medium, wherein when the computer program is executed by a processor, the computer is capable of performing the steps of the above-described method for selecting HLA typing tools based on typing tools to call intelligent agents and / or the above-described method for identifying susceptibility genes related to autoimmune diseases.

[0099] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the steps of the above-described HLA typing tool selection method based on typing tool invoking an agent and / or the above-described method for identifying susceptibility genes associated with autoimmune diseases.

[0100] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0101] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0102] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An HLA typing tool selection system based on typing tool invoking intelligent agents, characterized in that, include: The process coordination module is used to schedule the selection process of execution tools for various modules and intelligent agents within the system. The data acquisition module is used to acquire second-generation sequencing data of patients with specified diseases and reference sequencing data for calibrating or validating HLA typing results, based on sequencing conditions. The typing tool calls an intelligent agent to call several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data and obtain multiple HLA typing results. The performance evaluation module is used to determine the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in combination with reference sequencing data. The tool selection module is used to determine the preferred HLA typing tool based on the performance evaluation results.

2. The HLA typing tool selection system based on typing tool invoking intelligent agents according to claim 1, characterized in that, It also includes a tool performance database, which stores performance evaluation records of each HLA typing tool under different sequencing conditions and preferred HLA typing tool tags. The performance evaluation records are generated by the performance evaluation module and can be called by the tool selection module.

3. The HLA typing tool selection system based on typing tool invoking intelligent agents according to claim 1, characterized in that, Second-generation sequencing data includes any one or any combination of the following: whole exome sequencing data, MHC region targeted sequencing data, and RNA sequencing data; the reference sequencing data includes Sanger sequencing data.

4. The HLA typing tool selection system based on typing tool invoking intelligent agents according to claim 1, characterized in that, The HLA typing tool includes any one or any combination of the following: HISAT-genotype, OptiType, seq2HLA, HLA-HD, HLA LA, HLA-Ascan.

5. The HLA typing tool selection system based on typing tool invoking intelligent agents according to claim 3, characterized in that, The specified disease is an autoimmune disease, and the Sanger sequencing data includes sequencing data for four-position resolution typing of classic HLA class I and classic HLA class II loci for patients with autoimmune diseases.

6. The HLA typing tool selection system based on typing tool invoking intelligent agents according to any one of claims 1-4, characterized in that, The performance evaluation module is configured to evaluate the performance of the multiple HLA typing tools based on the multiple HLA typing results and in combination with reference sequencing data, using preset evaluation indicators to determine the performance evaluation results of each HLA typing tool. The preset evaluation indicators include any one of the following or any combination thereof: computational resource requirements, detection rate, accuracy, and robustness.

7. A method for selecting HLA typing tools based on calling an agent using a typing tool, characterized in that, include: Receive sequencing conditions input by the user; According to the sequencing conditions, obtain second-generation sequencing data of patients with the specified disease and reference sequencing data for calibrating or validating HLA typing results; The typing tool calls an agent to call several HLA typing tools to perform HLA typing analysis on the second-generation sequencing data, resulting in multiple HLA typing results; The performance evaluation module determines the performance evaluation results of each HLA typing tool based on the multiple HLA typing results and in conjunction with reference sequencing data. The tool selection module determines the preferred HLA typing tool based on the performance evaluation results.

8. A method for identifying susceptibility genes associated with autoimmune diseases, characterized in that, include: Acquire second-generation sequencing data of the target subject, wherein the target subject is a patient with an autoimmune disease; Analysis of sequencing conditions for second-generation sequencing data; Retrieve performance evaluation records and preferred HLA typing tool markers that match the sequencing conditions from the tool performance database to determine the preferred HLA typing tool. The performance evaluation records are generated by the HLA typing tool selection system based on the typing tool invoking agent according to any one of claims 1-6. The preferred HLA typing tool was used to perform HLA typing analysis on the second-generation sequencing data to obtain the target HLA typing results. Based on the target HLA typing results, an autoimmune disease association database is queried to identify susceptibility genes associated with the disease.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the program, it implements the HLA typing tool selection method for calling an agent based on typing tools as described in claim 7, and / or the method for identifying susceptibility genes related to autoimmune diseases as described in claim 8.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the HLA typing tool selection method based on typing tool invoking agent as described in claim 7, and / or the method for identifying susceptibility genes associated with autoimmune diseases as described in claim 8.

Images