Genome data variation detection method and system based on neural network
Through neural network -based genetic data variation detection methods and combined with the second test of the GATK tool, the accuracy and comprehensive insufficient comprehensiveness of mutation detection in existing technologies have been solved, and more efficient and accurate genome data variation detection is achieved.
Patent Information
- Application Number
- CN202411454244.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-17
- Publication Date
- 2025-01-24
AI Technical Summary
Existing genomic data analysis methods have information loss and inefficient processing when processing high-dimensional complex data. The traditional variation detection methods lack sensitivity and specificity in preliminary detection, making it difficult to fully capture all variant characteristics in the genome.
Using a neural network-based genomic data mutation detection method, by acquiring and preprocessing the genome data of insect-borne pathogens, extracting sequence characteristics, positional characteristics and environmental characteristics, training the neural network for preliminary mutation detection, and using the GATK tool for secondary mutation detection, Finally, the two results are merged.
It has significantly improved the accuracy and comprehensiveness of mutation detection, can more effectively capture and identify potential mutation characteristics in genomic data, reduce false positives and false negatives, and improve the overall efficiency and accuracy of detection results.
Smart Images
Figure CN119360953A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of genome data analysis, and in particular to a genome data variation detection method and system based on neural network. Background Art
[0002] In recent years, significant progress has been made in the genomic research of insect-borne pathogens, especially in the acquisition and analysis of pathogen genomic data. With the maturity of high-throughput sequencing technology, researchers can quickly obtain comprehensive genomic data of insect-borne pathogens, including viruses, bacteria, and protozoa. These technological advances have made the generation of genomic data more efficient and promoted in-depth research on the genetic variation of insect-borne pathogens and its relationship with pathogenicity. However, despite obvious technological advances, existing genomic data analysis methods still face many challenges, including the complexity of data processing and the accuracy of variation detection.
[0003] In the prior art, the analysis of genomic data usually relies on standard preprocessing and variant detection tools. However, these methods often have significant deficiencies when processing large amounts of complex data. First, existing data preprocessing and format conversion tools are not adaptable enough to processing high-dimensional data, especially when the data volume is large and the structure is complex, which may lead to information loss or inefficient processing. Secondly, although traditional variant detection methods such as the GATK tool perform well in terms of accuracy, the problems of sensitivity and specificity in preliminary variant detection have not been effectively solved. In addition, the existing single variant detection method is difficult to fully capture all the variant features in the genome, thereby limiting the comprehensive understanding and analysis of pathogen variants. Therefore, the prior art has certain limitations in the comprehensive processing, analysis and interpretation of insect-borne pathogen genomic data, especially the comprehensive efficiency and accuracy when multiple detection tools and methods are used in combination. Summary of the invention
[0004] In order to overcome the defects in the above-mentioned prior art, the present invention provides a method for detecting genome data variation based on a neural network, comprising: obtaining genome data of insect-borne pathogens, and preprocessing and format conversion of the genome data of the insect-borne pathogens; performing feature extraction on the genome data of the insect-borne pathogens after format conversion; training a neural network based on the extracted features, and performing preliminary variation detection using the trained neural network; performing secondary variation detection on the genome data of the insect-borne pathogens using the GATK tool, and fusing the preliminary variation detection results and the secondary variation detection results.
[0005] As a preferred solution of the neural network-based genome data variation detection method described in the present invention, the feature extraction includes extracting sequence characteristics, position characteristics and environmental characteristics; the sequence characteristics include: extracting k-mer frequency, variation information in the sequence context window; the position characteristics include: the position of the marker variation, the surrounding sequence background; the environmental characteristics include: extracting structural features in the genome.
[0006] As a preferred embodiment of the neural network-based genomic data variation detection method described in the present invention, the preprocessing includes data cleaning and data enhancement, and the format conversion includes mapping the preprocessed insect-borne pathogen genomic data into a matrix form.
[0007] As a preferred solution of the neural network-based genomic data variation detection method described in the present invention, the neural network includes an input layer, a feature extraction layer, a feature fusion layer, an attention mechanism layer, a detection head and an output layer, and adopts a SiLU activation function; the detection head includes three scales, namely bounding box regression, category prediction and confidence prediction.
[0008] As a preferred solution of the neural network-based genome data variation detection method of the present invention, it includes: dividing the extracted features into a training set and a test set, and normalizing the training set; initializing the weight W and learning rate α of the neural network, and constructing a loss function L:
[0009] L=(W / α)[YlogY-(1-y)log(1-y)]
[0010] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network;
[0011] Determine whether the loss function value is less than 1*10 -5 , if it is less than , the training ends;
[0012] Otherwise, the whale algorithm is used to find the optimal weights and obtain the optimal neural network.
[0013] As a preferred solution of the neural network-based genome data variation detection system described in the present invention, it includes: a data processing module, configured to obtain the genome data of the insect-borne pathogen, and pre-process and convert the format of the genome data of the insect-borne pathogen; a feature extraction module, configured to extract features from the format-converted genome data of the insect-borne pathogen; a first variation detection module, configured to train the neural network based on the extracted features, and perform preliminary variation detection using the trained neural network; a second variation detection module, configured to perform secondary variation detection on the genome data of the insect-borne pathogen using the GATK tool; and a variation detection module, configured to fuse the preliminary variation detection results and the secondary variation detection results.
[0014] As a preferred solution of the neural network-based genome data variation detection system described in the present invention, the feature extraction module is specifically configured to extract sequence characteristics, position characteristics and environmental characteristics; the sequence characteristics include: extracting k-mer frequency, variation information in the sequence context window; the position characteristics include: the position of the marker variation, the surrounding sequence background; the environmental characteristics include: extracting structural features in the genome.
[0015] As a preferred solution of the neural network-based genomic data variation detection system described in the present invention, the data processing module is specifically configured to perform: the preprocessing includes data cleaning and data enhancement, and the format conversion includes mapping the preprocessed insect-borne pathogen genomic data into a matrix form.
[0016] As a preferred solution of the neural network-based genomic data variation detection system described in the present invention, wherein: the first variation detection module is specifically configured to execute: the neural network includes an input layer, a feature extraction layer, a feature fusion layer, an attention mechanism layer, a detection head and an output layer, and adopts a SiLU activation function; the detection head includes three scales, namely bounding box regression, category prediction and confidence prediction.
[0017] As a preferred solution of the neural network-based genome data variation detection system of the present invention, the first variation detection module is specifically configured to execute: dividing the extracted features into a training set and a test set, and normalizing the training set; initializing the weights W and learning rate α of the neural network, and constructing a loss function L:
[0018] L=(W / α)[YlgY-(1-y)lg(1-y)]
[0019] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network;
[0020] Determine whether the loss function value is less than 5*10-7 , if it is less than , the training ends;
[0021] Otherwise, the whale algorithm is used to find the optimal weights and obtain the optimal neural network.
[0022] The beneficial effects of the present invention are as follows: by introducing a neural network for preliminary variation detection, it is possible to more effectively capture and identify potential variation features in genomic data, further utilize the GATK tool for secondary variation detection, and fuse the preliminary variation detection results with the secondary variation detection results, thereby significantly improving the accuracy and comprehensiveness of variation detection. BRIEF DESCRIPTION OF THE DRAWINGS
[0023] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative labor. Among them:
[0024] Figure 1 This is a schematic diagram of the flow of the neural network-based genome data variation detection method described in the first embodiment of the present invention. DETAILED DESCRIPTION
[0025] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.
[0026] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.
[0027] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor does it refer to a separate or selective embodiment that is mutually exclusive with other embodiments.
[0028] The present invention is described in detail with reference to schematic diagrams. When describing the embodiments of the present invention, for the sake of convenience, the cross-sectional diagrams showing the device structure will not be partially enlarged according to the general scale, and the schematic diagrams are only examples, which should not limit the scope of protection of the present invention. In addition, in actual production, the three-dimensional dimensions of length, width and depth should be included.
[0029] At the same time, in the description of the present invention, it should be noted that the directions or positional relationships indicated by the terms "upper, lower, inner and outer" are based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific direction, be constructed and operated in a specific direction, and therefore cannot be understood as limiting the present invention. In addition, the terms "first, second or third" are only used for descriptive purposes and cannot be understood as indicating or implying relative importance.
[0030] In the present invention, unless otherwise clearly specified and limited, the terms "install, connect, connect" should be understood in a broad sense, for example: it can be a fixed connection, a detachable connection or an integral connection; it can also be a mechanical connection, an electrical connection or a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.
[0031] Example 1
[0032] Reference Figure 1 , which is the first embodiment of the present invention, provides a method for detecting genome data variation based on a neural network, comprising:
[0033] S1: Obtain genomic data of insect-borne pathogens, and preprocess and convert the genomic data of insect-borne pathogens.
[0034] Automatically obtain genomic data of insect-borne pathogens from multiple external databases, including but not limited to NCBI (National Center for Biotechnology Information), GISAID (Global Initiative on Sharing Avian Influenza Data), etc.
[0035] Preprocessing includes data cleaning and data enhancement. Data cleaning is used to remove low-quality sequences and incorrect annotations to ensure data quality. Data enhancement uses MixUp and CutMix technologies. MixUp: MixUp enhances variant data to generate new samples by linearly combining images and labels of different samples. CutMix: Randomly cut and replace areas in the data to increase data diversity and help the model better identify variants.
[0036] Format conversion involves mapping the pre-processed vector-borne pathogen genomic data into a matrix form.
[0037] S2: Extract features from the format-converted insect-borne pathogen genome data.
[0038] Feature extraction includes extracting sequence characteristics, position characteristics and environmental characteristics;
[0039] Sequence features include: extracting k-mer frequency and variation information in the sequence context window;
[0040] Positional features include: the location of the marker variant and the surrounding sequence context;
[0041] Environmental characteristics include: extracting structural features in the genome, such as coding regions, non-coding regions, etc.
[0042] S3: Train the neural network based on the extracted features, and use the trained neural network to perform preliminary variation detection.
[0043] The neural network includes input layer, feature extraction layer, feature fusion layer, attention mechanism layer, detection head and output layer, and uses SiLU activation function;
[0044] The feature extraction layer includes CSPNet, convolution layer, batch normalization layer and Bottleneck layer, among which CSPNet (Cross-Stage Partial Network) is used to improve gradient flow and feature extraction, and Bottleneck layer is used to increase depth.
[0045] Feature fusion layer PANet and FPN, PANet (Path Aggregation Network) is used to achieve the fusion of features of different scales, and FPN is used to fuse multi-scale features to improve detection capabilities.
[0046] The input of the attention mechanism layer is connected to the output of the feature fusion layer to dynamically adjust the importance of feature channels and enhance feature representation capabilities.
[0047] The detection head consists of three scales, namely bounding box regression, category prediction, and confidence prediction.
[0048] The output layer is used to generate mutation prediction results, including the location, type, and confidence of the mutation.
[0049] Furthermore, in order to improve the prediction accuracy of the neural network, it is necessary to optimize the neural network training. Specifically, the extracted features are divided into a training set and a test set, and the training set is normalized;
[0050] Initialize the weights W and learning rate α of the neural network and construct the loss function L:
[0051] L=(W / α)[YlgY-(1-y)lg(1-y)]
[0052] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network;
[0053] Determine whether the loss function value is less than 5*10 -7 , if it is less than , the training ends;
[0054] Otherwise, the whale algorithm is used to find the optimal weights and obtain the optimal neural network.
[0055] The whale algorithm includes: taking the initial neural network weight W as the whale individual, initializing the number of whale individuals, the maximum number of iterations T and the number of neurons in the neural network;
[0056] Randomly generate the positions of individual whales and calculate the fitness of individual whales;
[0057] Update the position of the individual whales, calculate the fitness of the individual whales at this time, and select the best individual according to the fitness;
[0058] When the loss function value L of the neural network reaches the minimum and the maximum number of iterations T is reached, the training is stopped and the optimal weight is obtained;
[0059] Among them, the fitness of individual whales includes:
[0060] Fit=1 / L
[0061] Where Fit is the fitness of individual whales.
[0062] S4: Use GATK tools to perform secondary variation detection on insect-borne pathogen genomic data.
[0063] GATK (Genome Analysis Toolkit) is used to perform variant calls on the preprocessed insect-borne pathogen genomic data. Tools such as HaplotypeCaller are used. GATK identifies potential single nucleotide variants (SNPs) and insertion / deletion variants (Indels) based on the base information in the aligned data. The HaplotypeCaller tool constructs a local genotype model and comprehensively considers information such as base quality and candidate variant regions to more accurately predict variants.
[0064] Furthermore, the detected variants were filtered using the VariantFiltration tool to remove possible false positive results.
[0065] Furthermore, the variant results are recalibrated to improve the accuracy of variant detection. GATK uses a Gaussian mixture model to score variants and makes corrections based on the nature and quality of the variants.
[0066] S5: Fusion of the preliminary variation detection results and the secondary variation detection results.
[0067] The weighted average method is used to weight the initial mutation detection results and the secondary mutation detection results according to the prediction confidence.
[0068] Preferably, the variant results detected by the neural network are used as a preliminary screening and further verified by the GATK tool, which greatly reduces false positives and false negatives and improves the accuracy of the test results.
[0069] Example 2
[0070] This embodiment is different from the first embodiment in that it provides a genome data variation detection system based on a neural network, including:
[0071] The data processing module is configured to obtain the genomic data of insect-borne pathogens, and preprocess and convert the genomic data of insect-borne pathogens; the data processing module is specifically configured to perform: preprocessing includes data cleaning and data enhancement, data cleaning is used to remove low-quality sequences and incorrect annotations to ensure data quality; data enhancement uses MixUp and CutMix technologies, where MixUp: MixUp enhancement is performed on variant data to generate new samples by linearly combining images and labels of different samples; CutMix: randomly cut and replace regions in the data to increase data diversity and help the model better identify variants. Format conversion includes mapping the preprocessed genomic data of insect-borne pathogens into a matrix form.
[0072] The feature extraction module is configured to perform feature extraction on the format-converted insect-borne pathogen genome data; the feature extraction module is specifically configured to extract sequence characteristics, position characteristics and environmental characteristics; sequence characteristics include: extracting k-mer frequency, variation information in the sequence context window; position characteristics include: the position of the marker variation, the surrounding sequence background; environmental characteristics include: extracting structural features in the genome, such as coding regions, non-coding regions, etc.
[0073] The first variation detection module is configured to perform training of the neural network based on the extracted features, and use the trained neural network to perform preliminary variation detection; the first variation detection module is specifically configured to perform: the neural network includes an input layer, a feature extraction layer, a feature fusion layer, an attention mechanism layer, a detection head and an output layer, and uses a SiLU activation function; the feature extraction layer includes a CSPNet, a convolution layer, a batch normalization layer and a Bottleneck layer, wherein the CSPNet (Cross-Stage Partial Network) is used to improve the gradient flow and feature extraction, and the Bottleneck layer is used to increase the depth. The feature fusion layer PANet and FPN, PANet (Path Aggregation Network) is used to achieve the fusion of features of different scales, and FPN is used to fuse multi-scale features to improve the detection capability. The input of the attention mechanism layer is connected to the output of the feature fusion layer to dynamically adjust the importance of the feature channel and enhance the feature representation capability. The detection head includes three scales, namely bounding box regression, category prediction and confidence prediction. The output layer is used to generate variation prediction results, including the location, type and confidence of the variation.
[0074] Furthermore, in order to improve the prediction accuracy of the neural network, the neural network needs to be optimized and trained. Specifically, the extracted features are divided into a training set and a test set, and the training set is normalized; the weights W and learning rate α of the neural network are initialized, and the loss function L is constructed:
[0075] L=(W / α)[YlgY-(1-y)lg(1-y)]
[0076] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network;
[0077] Determine whether the loss function value is less than 5*10 -7 , if it is less than , the training ends;
[0078] Otherwise, the whale algorithm is used to find the optimal weights and obtain the optimal neural network.
[0079] The whale algorithm includes: taking the initial neural network weight W as the whale individual, initializing the number of whale individuals, the maximum number of iterations T and the number of neurons in the neural network;
[0080] Randomly generate the positions of individual whales and calculate the fitness of individual whales;
[0081] Update the position of the individual whales, calculate the fitness of the individual whales at this time, and select the best individual according to the fitness;
[0082] When the loss function value L of the neural network reaches the minimum and the maximum number of iterations T is reached, the training is stopped and the optimal weight is obtained;
[0083] Among them, the fitness of individual whales includes:
[0084] Fit=1 / L
[0085] Where Fit is the fitness of individual whales.
[0086] The second variant detection module is configured to perform secondary variant detection on the vector-borne pathogen genomic data using the GATK tool; use GATK (Genome Analysis Toolkit) to call variants on the preprocessed vector-borne pathogen genomic data, using tools such as HaplotypeCaller. GATK identifies potential single nucleotide variants (SNPs) and insertion / deletion variants (Indels) based on the base information in the alignment data. The HaplotypeCaller tool builds a local genotype model and comprehensively considers information such as base quality and candidate regions of variants to more accurately predict variants. Further, the detected variants are filtered using the VariantFiltration tool to remove possible false positive results. Further, the variant results are recalibrated to improve the accuracy of variant detection. GATK uses a Gaussian mixture model to score variants and calibrate them based on the nature and quality of the variants.
[0087] The variation detection module is configured to perform fusion of the preliminary variation detection result and the secondary variation detection result. The preliminary variation detection result and the secondary variation detection result are weightedly fused according to the prediction confidence using a weighted average method.
[0088] It should be appreciated that embodiments of the present invention may be implemented or enforced by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable memory. The method may be implemented in a computer program using standard programming techniques - including a non-transitory computer-readable storage medium configured with a computer program, wherein the storage medium so configured causes the computer to operate in a specific and predefined manner - according to the methods and drawings described in the specific embodiments. Each program may be implemented in a high-level procedural or object-oriented programming language to communicate with a computer system. However, if desired, the program may be implemented in an assembly or machine language. In any case, the language may be a compiled or interpreted language. In addition, the program may be run on a programmed ASIC for this purpose.
[0089] Furthermore, the operations of the processes described herein may be performed in any suitable order unless otherwise indicated herein or otherwise clearly contradicted by context. The processes described herein (or variations and / or combinations thereof) may be performed under the control of one or more computer systems configured with executable instructions, and may be implemented as code (e.g., executable instructions, one or more computer programs, or one or more applications) that is executed collectively on one or more processors, by hardware, or a combination thereof. The computer program includes a plurality of instructions that may be executed by one or more processors.
[0090] Further, the method can be implemented in any type of computing platform that is operably connected to a suitable computer, including but not limited to a personal computer, a minicomputer, a mainframe, a workstation, a network or distributed computing environment, a separate or integrated computer platform, or in communication with a charged particle tool or other imaging device, etc. Aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether removable or integrated into a computing platform, such as a hard disk, an optical read and / or write storage medium, a RAM, a ROM, etc., so that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the process described herein. In addition, the machine-readable code, or part thereof, can be transmitted via a wired or wireless network. When such media includes instructions or programs that implement the steps described above in conjunction with a microprocessor or other data processor, the invention described herein includes these and other different types of non-transitory computer-readable storage media. When programmed according to the methods and techniques described in the present invention, the present invention also includes the computer itself. The computer program can be applied to input data to perform the functions described herein, thereby converting the input data to generate output data stored in a non-volatile memory. The output information can also be applied to one or more output devices such as a display. In a preferred embodiment of the present invention, the converted data represents a physical and tangible object, including a specific visual depiction of the physical and tangible object produced on a display.
[0091] As used in this application, the terms "component", "module", "system", etc. are intended to refer to a computer-related entity, which can be hardware, firmware, a combination of hardware and software, software, or software in operation. For example, a component can be, but is not limited to: a process running on a processor, a processor, an object, an executable file, a thread in execution, a program, and / or a computer. As an example, an application running on a computing device and the computing device can both be components. One or more components can exist in a process and / or thread in execution, and a component can be located in a computer and / or distributed between two or more computers. In addition, these components can be executed from various computer-readable media having various data structures thereon. These components can communicate in a local and / or remote process manner, such as according to a signal having one or more data packets (e.g., data from a component that interacts with another component in a local system, a distributed system, and / or interacts with other systems in a signal manner through a network such as the Internet).
[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.
Claims
1. A method for detecting genomic data variation based on a neural network, characterized in that: include: Acquiring genomic data of insect-borne pathogens, and preprocessing and format-converting the genomic data of insect-borne pathogens; Perform feature extraction on the format-converted insect-borne pathogen genome data; The neural network is trained based on the extracted features, and the trained neural network is used for preliminary variation detection; The GATK tool was used to perform secondary variation detection on the genome data of insect-borne pathogens, and the primary variation detection results were fused with the secondary variation detection results.
2. The method for detecting genomic data variation based on a neural network according to claim 1, characterized in that: The feature extraction includes extracting sequence characteristics, position characteristics and environmental characteristics; Sequence features include: extracting k-mer frequency and variation information in the sequence context window; Positional features include: the location of the marker variant and the surrounding sequence context; Environmental features include: extracting structural features in the genome.
3. The method for detecting genomic data variation based on a neural network according to claim 1 or 2, characterized in that: The preprocessing includes data cleaning and data enhancement, and the format conversion includes mapping the preprocessed insect-borne pathogen genome data into a matrix form.
4. The method for detecting genomic data variation based on a neural network as claimed in claim 3, characterized in that: The neural network includes an input layer, a feature extraction layer, a feature fusion layer, an attention mechanism layer, a detection head and an output layer, and adopts a SiLU activation function; The detection head includes three scales, namely bounding box regression, category prediction and confidence prediction.
5. The method for detecting genomic data variation based on a neural network as claimed in claim 4, characterized in that: include: Dividing the extracted features into a training set and a test set, and performing normalization processing on the training set; Initialize the weights W and learning rate α of the neural network and construct the loss function L: L=(W / α)[YlogY-(1-y)log(1-y)] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network; Determine whether the loss function value is less than 1*10 -5 , if it is less than , the training ends; Otherwise, the optimal weights are found through the whale algorithm to obtain the optimal neural network.
6. A genomic data variation detection system based on neural network, characterized in that: include: A data processing module is configured to obtain genomic data of insect-borne pathogens, and perform preprocessing and format conversion on the genomic data of insect-borne pathogens; A feature extraction module is configured to perform feature extraction on the insect-borne pathogen genome data after format conversion; A first variation detection module is configured to perform training of a neural network based on the extracted features and perform preliminary variation detection using the trained neural network; A second variation detection module is configured to perform secondary variation detection on the insect-borne pathogen genome data using the GATK tool; The variation detection module is configured to perform fusion of the preliminary variation detection result and the secondary variation detection result.
7. The neural network-based genome data variation detection system according to claim 6, characterized in that: The feature extraction module is specifically configured to extract sequence characteristics, position characteristics and environmental characteristics; Sequence features include: extracting k-mer frequency and variation information in the sequence context window; Positional features include: the location of the marker variant and the surrounding sequence context; Environmental features include: extracting structural features in the genome.
8. The neural network-based genome data variation detection system according to claim 6 or 7, characterized in that: The data processing module is specifically configured to perform: the preprocessing includes data cleaning and data enhancement, and the format conversion includes mapping the preprocessed insect-borne pathogen genome data into a matrix form.
9. The neural network-based genome data variation detection system according to claim 8, characterized in that: The first variation detection module is specifically configured to execute: the neural network includes an input layer, a feature extraction layer, a feature fusion layer, an attention mechanism layer, a detection head and an output layer, and adopts a SiLU activation function; The detection head includes three scales, namely bounding box regression, category prediction and confidence prediction.
10. The neural network-based genome data variation detection system according to claim 9, characterized in that: The first variation detection module is specifically configured to perform: Dividing the extracted features into a training set and a test set, and performing normalization processing on the training set; Initialize the weights W and learning rate α of the neural network and construct the loss function L: L=(W / α)[YlgY-(1-y)lg(1-y)] In the formula, y is the actual output of the neural network, and Y is the predicted output of the neural network; Determine whether the loss function value is less than 5*10 -7 , if it is less than , the training ends; Otherwise, the optimal weights are found through the whale algorithm to obtain the optimal neural network.