Method for constructing a model for analyzing variant calling results
By constructing a random forest classification model based on GATK software and utilizing the feature parameters of variant sites, the problem of high false positive mutation rate in mutation detection was solved, achieving rapid and accurate mutation localization and cost reduction.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- TIANJIN HUADA GENE TECH CO LTD
- Filing Date
- 2022-04-25
- Publication Date
- 2026-05-12
AI Technical Summary
Existing technologies for mutation detection suffer from problems such as high false positive mutation recall rate, information loss, increased cost and turnaround time. Furthermore, existing models have fewer false positive mutations in their training sets, wider confidence intervals, and limited applicability or a high risk of overfitting.
A machine learning model is constructed, and feature parameters such as AD0, AD1, AF0, AF1, GT, DP, GQ, MQ, and QUAL values of variant sites are extracted using GATK software. A random forest classification model is used to detect variants, and a set of feature parameters is selected to build a model to predict false positive variants and determine genotypes.
Accurately predict false positive variants, reduce the cost and turnaround time of orthogonal experiments, and improve the accuracy and speed of variant localization.
Smart Images
Figure BDA0003614867330000031 
Figure BDA0003614867330000041 
Figure BDA0003614867330000081
Abstract
Description
Technical Field
[0001] This invention relates to the field of biology. Specifically, it relates to a method for constructing a model for analyzing variation detection results. Background Technology
[0002] Clinical next-generation sequencing (cNGS) is widely used to determine the molecular diagnosis of patients with genetic diseases. However, known NGS workflows are prone to random and systematic errors in sequencing, alignment, and variant recall steps. Because reported variants can impact patient care and treatment, the American College of Medical Genetics and Genomics (ACMG) and the American College of Pathologists (CAP) recommend orthogonal validation of reported variants to reduce the risk of errors and improve outcomes. Currently, Sanger sequencing remains the dominant technology for the molecular diagnosis of genetic diseases. However, as evidenced by the growth of public databases such as ClinVar and OMIM, the total number of clinically reported candidate variants is steadily increasing, multiplying testing costs and turnaround times, making complete sequencing increasingly impractical. Therefore, the need for machine learning models trained on large amounts of known data to identify false positive variants in cNGS data and reduce the reliance on orthogonal testing is becoming increasingly urgent.
[0003] Current research on variant false positives faces the following problems: orthogonal experiments such as Sanger sequencing increase costs and turnaround time significantly; existing models primarily use Boolean labels, which can lead to information loss compared to unchanged quantitative indicators; existing models have relatively few false positive variants in their training sets, potentially resulting in wider confidence intervals for certain false positive capture rates (especially SNVs); and due to cost considerations, existing models do not utilize sufficient clinical data, either being deliberately complex to suit multiple scenarios but lacking sufficient confidence, or having sufficient confidence but a high risk of overfitting and limited applicability.
[0004] Therefore, methods for predicting false positives for variants still need further research. Summary of the Invention
[0005] The present invention aims to solve at least one of the technical problems existing in the prior art.
[0006] Therefore, in one aspect of the present invention, a method for constructing a model for analyzing variant detection results is proposed. According to an embodiment of the present invention, the method includes: obtaining a positive sequencing dataset containing clearly identified positive variant sites and a negative sequencing dataset containing negative variant sites; extracting features of the variant sites from the positive and negative sequencing datasets respectively; and constructing a model using the feature results obtained in the previous step; wherein the features include at least one of the following: AD0 value: depth of the first allele in the variant site genotype; AD1 value: depth of the second allele in the variant site genotype; AF0 value: frequency of the first allele in the variant site genotype; AF1 value: frequency of the second allele in the variant site genotype; GT value: a single numerical value (specifically, 0, 1, 2, or 3); DP value: sequencing depth value; GQ value: quality value of the variant site genotype; MQ value: quality of the variant site mapping; and QUAL value: quality value of the variant site probability.
[0007] The variant detection and analysis software can generate dozens of feature parameters. The inventors compared and analyzed these feature parameters, selected a set of feature parameters, and used these feature parameters as attributes to build machine learning models for datasets with known positive and negative variant sites. The obtained model can accurately predict whether positive variant data are false positives, and can also further determine the genotype of the variant site, which helps to locate possible variants more quickly and accurately, and reduce the cost and turnaround time of orthogonal experiments.
[0008] In another aspect, the present invention proposes a method for analyzing variant detection results. According to an embodiment of the present invention, the method includes: acquiring a candidate positive variant dataset; and analyzing the candidate positive variant dataset using a machine learning model obtained through the aforementioned method for constructing a model for analyzing variant detection results, in order to predict whether the positive variant data in the candidate positive variant dataset are false positives and / or the genotype of the variant site. Thus, the method of the present invention can accurately predict whether the positive variant data are false positives, and can also determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0009] In another aspect, the present invention provides an apparatus for constructing a model for analyzing variant detection results. According to an embodiment of the present invention, the apparatus includes: an acquisition module adapted to acquire a positive sequencing dataset containing clearly identified positive variant sites and a negative sequencing dataset containing negative variant sites; an extraction module adapted to extract features of the variant sites from the positive and negative sequencing datasets, respectively; and a construction module adapted to construct a model using the feature results obtained by the extraction module; wherein the features include at least one of the following: AD0 value: depth of the first allele in the variant site genotype; AD1 value: depth of the second allele in the variant site genotype; AF0 value: frequency of the first allele in the variant site genotype; AF1 value: frequency of the second allele in the variant site genotype; GT value: single numerical value; DP value: sequencing depth value; GQ value: quality value of the variant site genotype; MQ value: quality of variant site mapping; and QUAL value: quality value of variant site probability. Therefore, the model obtained using the device of the present invention can accurately predict whether positive variant data are false positives, and can also determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0010] In another aspect, the present invention provides an executable storage medium. According to an embodiment of the invention, the storage medium stores computer program instructions that, when executed on a processor, cause the processor to perform the method for analyzing mutation detection results as described above. Thus, by executing the storage medium of the present invention, it is possible to accurately predict whether positive mutation data are false positives, and also to determine the genotype of the mutation, which helps to locate potential mutations more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0011] In another aspect, the present invention provides an electronic device. According to an embodiment of the invention, the electronic device includes: the aforementioned executable storage medium; and the processor for executing the computer program to implement the aforementioned method for analyzing variant detection results. Thus, by implementing the electronic device of the present invention, it is possible to accurately predict whether positive variant data are false positives, and also to determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0012] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Detailed Implementation
[0013] The embodiments of the present invention are described in detail below. These embodiments are exemplary and are only used to explain the present invention, and should not be construed as limiting the invention. Where specific techniques or conditions are not specified in the embodiments, they are performed according to the techniques or conditions described in the literature in the art or according to the product instructions. Reagents or instruments used, unless otherwise specified, are all commercially available conventional products.
[0014] Methods for constructing models for analyzing variant detection results
[0015] In one aspect of the present invention, a method for constructing a model for analyzing variant detection results is proposed. According to an embodiment of the present invention, the method includes: obtaining a positive sequencing dataset containing clearly identified positive variant sites and a negative sequencing dataset containing negative variant sites; extracting features of the variant sites from the positive and negative sequencing datasets, respectively; and constructing a model using the feature results obtained in the previous step; wherein the features include at least one of the following: AD0 value, AD1 value, AF0 value, AF1 value, GT value, DP value, GQ value, MQ value, and QUAL value.
[0016] The inventors, through extensive experimentation, selected the aforementioned nine feature parameters, all of which are feature parameters from the GATK software. Their specific meanings are shown in the table below. Using these as feature attributes, machine learning was applied to positive sequencing datasets containing clearly identified positive variant sites and negative sequencing datasets containing negative variant sites to obtain a predictive model. Therefore, the obtained model can accurately predict whether a positive variant data point is a false positive, helping to locate potential variants more quickly and accurately, and reducing the cost and turnaround time of orthogonal experiments.
[0017] Table 1. Meaning of Features
[0018]
[0019]
[0020] According to an embodiment of the present invention, the positive sequencing dataset with clearly identified positive variant sites and the negative sequencing dataset with clearly identified negative variant sites are obtained by the following method: obtaining a sequencing dataset; comparing the sequencing dataset with reference data using GATK software to obtain a candidate positive variant dataset; and analyzing the candidate positive variant dataset to obtain a positive sequencing dataset with clearly identified positive variant sites and a negative sequencing dataset with clearly identified negative variant sites.
[0021] The first step is to acquire clinical gene sequencing data. This data is then compared with reference data (including alignment, variant detection, annotation, and filtering), and variants are identified using GATK to obtain candidate positive variant data, which is then output as a VCF file. The candidate positive variant data is further analyzed to determine whether the data are true positives or false positives. The data is then divided into a positive sequencing dataset for positive variant sites and a negative sequencing dataset for negative variant sites.
[0022] According to an embodiment of the present invention, the reference sequence is selected from the human genome hg19.
[0023] According to an embodiment of the present invention, the analysis process includes: performing standard clinical interpretation on the candidate positive variant dataset to obtain a potentially pathogenic variant dataset; performing orthogonal experimental analysis on the potentially pathogenic variant dataset to obtain a positive sequencing dataset with clearly identified positive variant sites and a negative sequencing dataset with negative variant sites, wherein the positive sequencing dataset includes an SNV variant type dataset and an INDEL variant type dataset, and the SNV variant type dataset and the INDEL variant type dataset respectively include a homozygous genotype dataset and a heterozygous genotype dataset.
[0024] The term "standard clinical interpretation" refers to the interpretation of the pathogenicity of clinical variants with reference to the 2015 ACMG guidelines.
[0025] By performing standard clinical interpretation on candidate positive variant data obtained from GATK identification and analysis, potentially pathogenic variant data can be obtained. Then, orthogonal experiments are used to verify the accuracy of these variants, resulting in positive sequencing datasets clearly identifying positive variant sites and negative sequencing datasets identifying negative variant sites. Positive sequencing datasets can be categorized into SNV variant types and INDEL variant types. These two variant types can further accurately determine the genotype of the variant, i.e., homozygous (Hom) or heterozygous (Het).
[0026] It should be noted that the present invention does not strictly limit the method of orthogonal experimental analysis. As long as it is possible to determine whether the potentially pathogenic variant data is a true positive variant or a false positive, conventional techniques in the field can be used. For example, refer to Sanger F. DNA sequencing with chain-terminating inhibitors. 1977[J]. Biotechnology (Reading, Mass.), 24:104-108.
[0027] According to an embodiment of the present invention, the model is selected from a random forest classification model, and the threshold is 0.95±0.05. The threshold setting ensures sufficient accuracy and reduces random errors. By employing a scalable threshold setting, a trade-off can be struck between accuracy and the rate of orthogonal experimentation while ensuring sufficient accuracy.
[0028] According to a specific embodiment of the present invention, the positive sequencing dataset with clearly identified positive variant sites and the negative sequencing dataset with negative variant sites are respectively divided into training set and test set (3:1), and a random forest classification model is selected. The model with the highest accuracy is selected through 5-fold cross-validation.
[0029] According to an embodiment of the present invention, the method for constructing the model for analyzing mutation detection results includes:
[0030] 1. First, obtain clinical genomic data, align it with the human reference genome (hg19), and use GATK to identify variants and output a VCF file;
[0031] 2. The variants that may cause disease are obtained through standard clinical interpretation, and the accuracy of the variants is verified through orthogonal experiments. The accurate genotypes Hom (homozygous), Het (heterozygous), and N (no variant) are provided.
[0032] 3. Then, the VCF file is converted into machine learning labels and features, from which a total of 9 features are obtained, as detailed in Table 1:
[0033] 4. Based on the different mutation types (SNV, INDEL), two different machine learning classification models are constructed using features extracted from the VCF file, and the optimal parameters are sought through grid search.
[0034] 5. Based on the above method, the data is divided into training set and test set (3:1), and a random forest classification model is selected. After 5-fold cross-validation, the model with the highest accuracy is selected.
[0035] Methods for analyzing variant detection results
[0036] In another aspect, the present invention proposes a method for analyzing variant detection results. According to an embodiment of the present invention, the method includes: acquiring a candidate positive variant dataset; and analyzing the candidate positive variant dataset using a machine learning model obtained through the aforementioned method for constructing a model for analyzing variant detection results, in order to predict whether the positive variant data in the candidate positive variant dataset are false positives and / or the genotype of the variant site. Thus, the model obtained using the method of the present invention can accurately predict whether the candidate positive variant data are false positives, and can also determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0037] According to an embodiment of the present invention, the candidate positive variant dataset is obtained by: acquiring a sequencing dataset; and using GATK software to compare the sequencing dataset with reference data to obtain the candidate positive variant dataset.
[0038] According to an embodiment of the present invention, the model is selected from a random forest classification model. When the confidence level of the candidate positive variant data is lower than the threshold of the model, the candidate positive variant data is subjected to orthogonal experimental analysis to predict whether the positive variant data in the candidate positive variant dataset is a false positive. Data below the threshold is called gray zone data. The accuracy of predicting false positives using the model is low. Therefore, it is necessary to further verify this part of the data with orthogonal experiments to accurately predict its false positives.
[0039] Those skilled in the art will understand that the features and advantages described above for the construction method of the model used to analyze the results of variant detection are also applicable to this method for analyzing the results of variant detection, and will not be repeated here.
[0040] A device for constructing models to analyze variation detection results.
[0041] In another aspect, the present invention provides an apparatus for constructing a model for analyzing variant detection results. According to an embodiment of the present invention, the apparatus includes: an acquisition module adapted to acquire a positive sequencing dataset containing clearly identified positive variant sites and a negative sequencing dataset containing negative variant sites; an extraction module adapted to extract features of the variant sites from the positive and negative sequencing datasets, respectively; and a construction module adapted to construct a model using the feature results obtained by the extraction module; wherein the features include at least one of the following: AD0 value: depth of the first allele in the variant site genotype; AD1 value: depth of the second allele in the variant site genotype; AF0 value: frequency of the first allele in the variant site genotype; AF1 value: frequency of the second allele in the variant site genotype; GT value: single numerical value; DP value: sequencing depth value; GQ value: quality value of the variant site genotype; MQ value: quality of variant site mapping; and QUAL value: quality value of variant site probability. Therefore, the model obtained using the device of the present invention can accurately predict whether positive variant data are false positives, and can also determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0042] According to an embodiment of the present invention, the acquisition module includes: a sequencing dataset acquisition module, which is adapted to acquire a sequencing dataset; a comparison processing module, which is adapted to compare the sequencing dataset with reference data using GATK software to obtain a candidate positive variant dataset; and an analysis processing module, which is adapted to analyze the candidate positive variant dataset to obtain a positive sequencing dataset with clearly identified positive variant sites and a negative sequencing dataset with negative variant sites. Using the acquisition module, the positive and negative variant site data in the sequencing dataset can be accurately determined, and the genotype of the positive variant sites can also be determined.
[0043] According to an embodiment of the present invention, the analysis and processing module includes: a standard clinical interpretation module, which is adapted to perform standard clinical interpretation on the positive variant data to obtain potentially pathogenic variant data; and an orthogonal experiment analysis module, which is adapted to perform orthogonal experiment analysis on the potentially pathogenic variant data to obtain a positive sequencing dataset with clearly identified positive variant sites and a negative sequencing dataset with negative variant sites.
[0044] Executable storage medium
[0045] In another aspect, the present invention provides an executable storage medium. According to an embodiment of the invention, the storage medium stores computer program instructions that, when executed on a processor, cause the processor to perform the method for analyzing variant detection results as described above. Thus, by executing the storage medium of the present invention, it is possible to accurately predict whether positive variant data are false positives, and simultaneously determine the genotype of the variant, facilitating faster and more accurate localization of potential variants and reducing the cost and turnaround time of orthogonal experiments.
[0046] Those skilled in the art will understand that the features and advantages described above for the method of analyzing mutation detection results also apply to this executable storage medium, and will not be repeated here.
[0047] electronic devices
[0048] In another aspect, the present invention provides an electronic device. According to an embodiment of the invention, the electronic device includes: the aforementioned executable storage medium; and a processor for executing the computer program to implement the aforementioned method for analyzing variant detection results. Thus, by implementing the electronic device of the present invention, it is possible to accurately predict whether positive variant data are false positives, and also to determine the genotype of the variant, which helps to locate possible variants more quickly and accurately, and reduces the cost and turnaround time of orthogonal experiments.
[0049] Those skilled in the art will understand that the features and advantages described above regarding the method for analyzing mutation detection results and the executable storage medium are equally applicable to this electronic device, and will not be repeated here.
[0050] The present invention will be explained below with reference to embodiments. Those skilled in the art will understand that the following embodiments are for illustrative purposes only and should not be considered as limiting the scope of the invention. Where specific techniques or conditions are not specified in the embodiments, they are performed according to the techniques or conditions described in the literature in the field or according to the product instructions. Reagents or instruments whose manufacturers are not specified are all conventional products that can be obtained commercially.
[0051] Example 1
[0052] 1. We obtained WES data from 5190 clinical patients and used GATK software to compare the data with the human genome hg19, perform variant detection, annotation, and filtering to obtain VCF files;
[0053] 2. The VCF file was analyzed using a standard clinical interpretation process, resulting in 7375 potentially pathogenic variants.
[0054] 3. Orthogonal experiments were conducted to verify the 7375 variants mentioned above (see Sanger F. DNA sequencing with chain-terminating inhibitors. 1977[J]. Biotechnology (Reading, Mass.), 24:104-108 for details). These variants were determined to include 5241 variant types (SNV) and 2134 variant types (INDEL). Among the SNVs, there were 3226 Het genotypes, 63 Hom genotypes, and 1952 negative variants; among the INDELs, there were 1606 Het genotypes, 138 Hom genotypes, and 390 negative variants.
[0055] 4. Divide the data from the previous step into a training set and a test set (3:1). Build a random forest classification model for the training set. Use all the features in the training set as candidate features and then perform principal component analysis. Finally, the 9 features listed in Table 2 were determined.
[0056] Table 2. Feature Importance in Random Forest Classification Models Based on Different Variance Types (SNV and INDEL) Table 2. Feature Importance in Random Forest Classification Models Based on Different Variance Types (SNV and INDEL)
[0057] feature SNV_MODEL INDEL_MODEL AD0 0.0305 0.0389 AD1 0.0365 0.0606 AF0 0.3350 0.3135 AF1 0.2352 0.3027 GT 0.0078 0.0289 DP 0.0300 0.0174 GQ 0.0787 0.0691 MQ 0.0139 0.0141 QUAL 0.2324 0.1548
[0058] The test set accuracies of the SNV and INDEL models were 94.8% and 93.8%, respectively, with the accuracies for different genotypes shown in Table 3.
[0059] Table 3. Accuracy of different genotypes in the random forest classification model using different variant types SNV and INDEL.
[0060] genotype SNV_MODEL(%) INDEL_MODEL(%) Het 92.9 80.5 Hom 100 92.1 N (negative) 96.3 97.2
[0061] Considering the accuracy requirements of clinical data, this method uses different thresholds (confidence levels of random forest results) to obtain different accuracies and orthogonal experiment proportions on the test set (Table 4). Accuracy is defined as the number of correct judgments divided by the total number of results meeting the threshold, and the orthogonal experiment proportion is defined as the number of results below the threshold divided by the total number of test samples. A threshold with the smallest possible orthogonal experiment proportion, while maintaining sufficient accuracy, was chosen as the target threshold, ultimately set at 0.95, within a scalable range of ±0.05. The results show that this method has a certain tolerance for noisy data, data redundancy, and low-quality data, demonstrating good robustness.
[0062] Table 4. Different thresholds and the proportion of orthogonal experiments required for establishing random forest classification models for different mutation types SNV and INDEL.
[0063]
[0064]
[0065] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.
[0066] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention.
Claims
1. A method for constructing a model for analyzing mutation detection results, characterized in that, include: Obtain positive sequencing datasets for clearly identified positive variant sites and negative sequencing datasets for negative variant sites; Features of variant sites were extracted from the positive and negative sequencing datasets, respectively. Build a model using the feature results obtained in the previous step; The features include at least one of the following: AD0 value: The depth of the first allele in the genotype of the variant site; AD1 value: The depth of the second allele in the genotype of the variant site; AF0 value: the frequency of the first allele in the genotype at the variant site; AF1 value: the frequency of the second allele in the genotype at the variant site; GT value: a single numerical value; DP value: Sequencing depth value; GQ value: Quality value of the genotype at the mutation site; MQ value: Quality of variant site mapping; QUAL value: A quality value indicating the probability of a variant site; The positive sequencing datasets for clearly identified positive variant sites and the negative sequencing datasets for negative variant sites were obtained through the following methods: Obtain the sequencing dataset; The sequencing dataset was compared with the reference data using GATK software to obtain a candidate positive variant dataset. The candidate positive variant datasets are analyzed and processed to obtain positive sequencing datasets with clearly identified positive variant sites and negative sequencing datasets with negative variant sites. The analysis and processing includes: The candidate positive variant dataset is interpreted using standard clinical methods to obtain a dataset of potentially pathogenic variants. Orthogonal experimental analysis was performed on the potentially pathogenic variant dataset to obtain positive sequencing datasets with clearly identified positive variant sites and negative sequencing datasets with negative variant sites. The positive sequencing datasets include SNV variant type datasets and INDEL variant type datasets, which respectively include homozygous genotype datasets and heterozygous genotype datasets. The model is selected from the random forest classification model.
2. The method according to claim 1, characterized in that, The reference sequence was selected from the human genome hg19.
3. The method according to claim 1, characterized in that, The threshold of the random forest classification model is 0.95±0.
05.
4. A method for analyzing variation detection results, characterized in that, include: Obtain the candidate positive variant dataset; The machine learning model obtained by the model construction method for analyzing variant detection results according to any one of claims 1 to 3 is used to analyze the candidate positive variant dataset in order to predict whether the positive variant data in the candidate positive variant dataset are false positives and / or the genotype of the variant site.
5. The method according to claim 4, characterized in that, When the confidence level of the candidate positive variant data is lower than the threshold of the model, the candidate positive variant data is subjected to orthogonal experimental analysis in order to predict whether the positive variant data in the candidate positive variant dataset is a false positive.
6. An apparatus for constructing a model for analyzing mutation detection results, characterized in that, include: The acquisition module is adapted to acquire positive sequencing datasets that are clearly identified as positive variant sites and negative sequencing datasets that are identified as negative variant sites; An extraction module, adapted to extract features of variant sites from the positive sequencing dataset and the negative sequencing dataset, respectively; A construction module, wherein the construction module is adapted to construct a model using the feature results obtained by the extraction module; The features include at least one of the following: AD0 value: The depth of the first allele in the genotype of the variant site; AD1 value: The depth of the second allele in the genotype of the variant site; AF0 value: the frequency of the first allele in the genotype at the variant site; AF1 value: the frequency of the second allele in the genotype at the variant site; GT value: a single numerical value; DP value: Sequencing depth value; GQ value: Quality value of the genotype at the mutation site; MQ value: Quality of variant site mapping; QUAL value: A quality value indicating the probability of a variant site; The acquisition module includes: A module for acquiring sequencing datasets, wherein the module for acquiring sequencing datasets is adapted to acquire sequencing datasets; The comparison processing module is adapted to use GATK software to compare the sequencing dataset with reference data to obtain a candidate positive variant dataset. An analysis and processing module is provided, which is adapted to analyze and process the candidate positive variant dataset to obtain a positive sequencing dataset with clearly identified positive variant sites and a negative sequencing dataset with negative variant sites. The analysis and processing module includes: A standard clinical interpretation module, which is adapted to perform standard clinical interpretation on the positive variant data to obtain potentially pathogenic variant data; The orthogonal experiment analysis module is adapted to perform orthogonal experiment analysis on the potentially pathogenic variant data to obtain positive sequencing datasets with clearly identified positive variant sites and negative sequencing datasets with negative variant sites. The model is selected from the random forest classification model.
7. An executable storage medium, characterized in that, The storage medium stores computer program instructions that, when executed on a processor, cause the processor to perform the method for analyzing mutation detection results as described in claim 4 or 5.
8. An electronic device, characterized in that, include: The executable storage medium as described in claim 7; The processor is configured to execute the computer program to implement the method for analyzing mutation detection results as described in claim 4 or 5.