Method and device for processing multimodal brain imaging and genetic data

Abstract

Embodiments of the present invention disclose a method and apparatus for processing multimodal brain imaging and genetic data. This method includes using metabolic pathways to partition co-expressed gene sets and integrating multimodal brain imaging for association analysis. Specifically, a model is used to automatically derive target brain regions and their gene loci from the collected and organized data. This invention addresses the challenges of related art, such as difficulty in acquiring data and poor processing results, associated with association analysis based on raw genetic data and single-modality brain imaging data. This method achieves the technical advantage of reducing the difficulty of data acquisition and improving the accuracy of prediction results.

Description

Technical Field

[0001] The present invention relates to the technical field of digital image analysis and pattern recognition, and in particular to a method and device for processing multimodal brain images and genetic data. Background Art

[0002] Multimodal brain imaging-gene association analysis is a general term for methods in which researchers use brain imaging data from different modalities and genetic data to study brain regions and variant sites associated with complex diseases. Its goal is to discover brain imaging regions and gene loci associated with the disease.

[0003] Existing multimodal brain imaging-gene association analysis methods all utilize raw brain imaging data and genetic data from different modalities for analysis and validation. Raw brain imaging data includes functional magnetic resonance imaging (fMRI), structural magnetic resonance imaging (sMRI), diffusion tensor imaging (DTI), and more. Raw genetic data mostly consists of single nucleotide polymorphism (SNP) results. The primary characteristic of these technical methods is the collection and preprocessing of brain imaging and genetic data. Only on this basis can image-gene association analysis be performed. However, due to the confidentiality of raw data, access to these samples is very limited, making it extremely difficult to obtain large amounts of data results.

[0004] Secondly, existing association research methods typically employ independent and pairwise univariate analyses, treating genes or SNPs as independent units while ignoring important potential interactions between these units. For example, when considering the impact of SNPs on disease, they are constrained solely by a single norm, thus ignoring the interactions between units within the SNP. However, association analysis using only a single modality of brain imaging data is insufficient.

[0005] Regarding the correlation analysis based on original genetic data and single-modality brain imaging data in related technologies, there are problems such as difficulty in obtaining data and poor processing effects, and no effective solution has been proposed yet. Summary of the Invention

[0006] The embodiments of the present invention provide a method and device for processing multimodal brain images and genetic data, so as to at least solve the technical problems in the related art of performing association analysis based on original genetic data and single-modality brain imaging data, which have the disadvantages of difficulty in obtaining data and poor processing effect.

[0007] According to one aspect of an embodiment of the present invention, a method for processing multimodal brain imaging and genetic data is provided, comprising: obtaining genome-wide association study (GWAS) data of multimodal brain imaging and genetic data of a corresponding reference population; establishing a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix, and a gene-gene association matrix based on the GWAS data and the genetic data, respectively; performing pathway metabolic pathway analysis on the genetic data, and grouping co-expressed genes according to the analysis results to obtain a co-expressed gene set; and processing the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix using a multimodal brain imaging-gene association model based on the co-expressed gene set to obtain a target brain region and the gene loci of the target brain region.

[0008] Optionally, obtaining genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population includes: obtaining the GWAS data of the multimodal brain imaging; and obtaining the genetic data corresponding to the reference population based on the gene sequence in the GWAS data.

[0009] Optionally, based on the GWAS data and the gene data, a gene-brain image association matrix, a brain image-brain image association matrix and a gene-gene association matrix are respectively established, including: generating the gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; transforming the gene-brain image association matrix to obtain the brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; generating a gene-gene association matrix based on the gene data and the gene sequence in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p-dimensional matrix; wherein p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain image.

[0010] Optionally, before generating the gene-brain imaging association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, the method further includes: determining whether the gene data and the brain imaging data are normalized before performing GWAS; if the gene data and the brain imaging data are not normalized, correcting the weight values ​​corresponding to the GWAS data.

[0011] Optionally, the weight values ​​corresponding to the GWAS data are corrected using the following correction formula: Among them, β STANDRrepresents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

[0012] Optionally, based on the co-expressed gene set, the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are processed using a multimodal brain image-gene association model to obtain the target brain region and the gene loci of the target brain region, including: determining the objective function of the multimodal brain image-gene association model, wherein the expression of the objective function is as follows:

[0013]

[0014] in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the characteristic dimension of the gene, q represents the characteristic dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modes, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges to obtain the target brain region and the gene loci of the target brain region.

[0015] Optionally, the gene-brain image association matrix, the brain image-brain image association matrix and the gene-gene association matrix are respectively input into the objective function for processing, comprising: inputting λ1, ..., λ m ,γ,α,τ1,...,τ m ,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialize u∈R p×1 and v1,...,v m ∈R q ×1 ; Update D G , where D G is a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; solve for u so that Update D m , where D m is a diagonal matrix, each diagonal element is Solve for v1,...,v m , and make

[0016] According to another aspect of an embodiment of the present invention, a device for processing multimodal brain images and genetic data is also provided, including: an acquisition module for acquiring genome-wide association analysis GWAS data of multimodal brain images and genetic data of a corresponding reference population; an establishment module for establishing a gene-brain image association matrix, a brain image-brain image association matrix, and a gene-gene association matrix based on the GWAS data and the genetic data, respectively; a grouping module for performing pathway metabolic pathway analysis on the genetic data, and grouping co-expressed genes according to the analysis results to obtain a co-expressed gene set; a processing module for processing the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix using a multimodal brain image-gene association model based on the co-expressed gene set to obtain a target brain region and the gene loci of the target brain region.

[0017] According to another aspect of an embodiment of the present invention, an electronic device is also provided, comprising: a processor; and a memory for storing processor-executable instructions; wherein the processor is configured to execute any one of the above-mentioned methods for processing multimodal brain imaging and genetic data.

[0018] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned methods for processing multimodal brain images and genetic data.

[0019] In an embodiment of the present invention, genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population are obtained; based on the GWAS data and genetic data, a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix, and a gene-gene association matrix are respectively established; pathway metabolic pathway analysis is performed on the genetic data, and co-expressed genes are grouped according to the analysis results to obtain a co-expressed gene set; based on the co-expressed gene set, a multimodal brain imaging-gene association model is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix to obtain the gene loci of the target brain region and the target brain region. In other words, the embodiment of the present invention uses metabolic pathways to divide the co-expressed gene set and conducts association analysis based on multimodal brain imaging. Specifically, through the collected and organized data, the model is used to automatically derive the gene loci of the target brain region and the target brain region, thereby solving the problem of difficulty in obtaining data and poor processing effect in the related art of association analysis based on original genetic data and single-modality brain imaging data, thereby achieving the technical effect of reducing the difficulty of data acquisition and improving the accuracy of prediction results. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of this application. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0021] Figure 1 is a flow chart of a method for processing multimodal brain imaging and genetic data according to an embodiment of the present invention;

[0022] Figure 2 is a flow chart of another method for processing multimodal brain imaging and genetic data according to an embodiment of the present invention;

[0023] Figure 3 2 is a schematic diagram of a multimodal brain image and gene data processing device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0024] In order to enable those skilled in the art to better understand the solutions of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts should fall within the scope of protection of the present invention.

[0025] It should be noted that the terms "first", "second", etc. in the specification, claims and drawings of the present invention are used to distinguish different objects rather than to limit a specific order.

[0026] For the convenience of description, some nouns or terms appearing in the present invention are described in detail below.

[0027] Genome-wide association analysis (GWAS) is a univariate regression analysis that can reflect the association between a single modality of brain imaging and a single gene. GWAS analysis data results are widely available and are easily accessible compared to raw multimodal brain imaging and genetic data. Therefore, the present invention no longer relies on raw data acquisition, but instead collects existing GWAS analysis results from different modalities.

[0028] Multimodal imaging refers to the use of brain imaging data from multiple different imaging technologies in a single computational model, or different indicators under the same imaging technology, such as gray matter density, cortical thickness, cerebrospinal fluid, etc. The tissue structures of different modalities provide different brain information. This invention collects and organizes GWAS analysis results of different modalities, explores the association between different modal brain images and genes, and ultimately can discover a richer range of disease-related brain lesions.

[0029] An embodiment of the present invention provides a method for processing multimodal brain imaging and genetic data. It should be noted that the steps shown in the flowchart of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and although a logical order is shown in the flowchart, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0030] Figure 1 FIG. 1 is a flow chart of a method for processing multimodal brain imaging and genetic data according to an embodiment of the present invention. Figure 1 As shown, the method includes the following steps:

[0031] Step S102, obtaining genome-wide association analysis (GWAS) data of multimodal brain imaging and genetic data of corresponding reference populations;

[0032] Step S104, establishing a gene-brain image association matrix, a brain image-brain image association matrix, and a gene-gene association matrix based on the GWAS data and the gene data;

[0033] Step S106, performing pathway metabolic pathway analysis on the gene data, and grouping the co-expressed genes according to the analysis results to obtain a co-expressed gene set;

[0034] It should be noted that the SNPs used can be grouped according to gene categories, and the grouped gene pathways can be analyzed. After the analysis is completed, the corresponding genes can be grouped according to the pathway analysis results. The corresponding weight constraints are:

[0035]

[0036] Among them, K represents the number of groups, and μ represents each SNP in the gene.

[0037] Step S108 , based on the co-expressed gene set, the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are processed using a multimodal brain image-gene association model to obtain the target brain region and the gene loci of the target brain region.

[0038] These target brain regions include, but are not limited to, disease-related lesions. Furthermore, different modalities provide distinct brain information, and data from different modalities can complement and integrate each other. Therefore, integrating multimodal brain imaging can reveal richer disease-related phenotypic features.

[0039] The present invention uses pathway metabolic pathway analysis to find multiple groups of co-expressed gene sets by analyzing gene enrichment pathways. Then, by grouping the co-expressed gene sets and using the association analysis model of the present invention, the gene characteristic sites related to the diseased brain area are finally found.

[0040] It should be noted that this method changes the approach to data collection and preprocessing. Rather than using the original brain imaging and genetic data, it instead collects and uses genome-wide association study data as the sample data for the association analysis. This significantly reduces the difficulty of data acquisition due to the availability and accessibility of GWAS data.

[0041] Secondly, complex diseases are often not caused by a single gene or single SNP. Instead, multiple genes are involved in the pathogenesis of the disease, interacting and influencing each other to form a network. Studying the mechanisms and genetic basis of disease from the perspective of the combined effects of multiple genes is a more accurate and effective approach.

[0042] Finally, using only a single modality of brain imaging data for correlation analysis is insufficient. Different modalities provide distinct brain structures, and data from different modalities can complement and integrate each other. Therefore, integrating multimodal brain imaging can reveal richer disease-related phenotypic features.

[0043] In an embodiment of the present invention, genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population are obtained; based on the GWAS data and genetic data, a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix, and a gene-gene association matrix are respectively established; pathway metabolic pathway analysis is performed on the genetic data, and co-expressed genes are grouped according to the analysis results to obtain a co-expressed gene set; based on the co-expressed gene set, a multimodal brain imaging-gene association model is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix to obtain the gene loci of the target brain region and the target brain region. In other words, the embodiment of the present invention uses metabolic pathways to divide the co-expressed gene set and conducts association analysis based on multimodal brain imaging. Specifically, through the collected and organized data, the model is used to automatically derive the gene loci of the target brain region and the target brain region, thereby solving the problem of difficulty in obtaining data and poor processing effect in the related art of association analysis based on original genetic data and single-modality brain imaging data, thereby achieving the technical effect of reducing the difficulty of data acquisition and improving the accuracy of prediction results.

[0044] In an optional embodiment, obtaining genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population includes: obtaining GWAS data of multimodal brain imaging; and obtaining genetic data of a corresponding reference population based on gene sequences in the GWAS data.

[0045] Alternatively, you can collect genome-wide association analysis data related to multimodal brain imaging and check whether the beta values ​​in the GWAS results are normalized. The 1000 Genomes Project is the most comprehensive and medically valuable map of human genetic polymorphisms to date. Therefore, based on the population characteristics of the different GWAS data collected, you can download the genetic data of the corresponding reference population from the 1000 Genomes Project.

[0046] In an optional embodiment, a gene-brain image association matrix, a brain image-brain image association matrix and a gene-gene association matrix are established respectively based on the GWAS data and the genetic data, including: generating a gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; transforming the gene-brain image association matrix to obtain a brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; generating a gene-gene association matrix based on the genetic data and the gene sequence in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p-dimensional matrix; wherein p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain image.

[0047] Gene-brain imaging correlation matrix: using XYm represents the gene-brain imaging association matrix. m represents the number of brain imaging modalities. The β values ​​from the GWAS data of the same modality are integrated into a p × q dimensional matrix. Here, p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain imaging.

[0048] Brain image-brain image correlation matrix: use Y m Y m Represents the brain image-brain image correlation matrix. Based on the gene-brain image correlation matrix, it is converted into a brain image-brain image correlation matrix using the corresponding formula. The specific algorithm formula is as follows:

[0049]

[0050] Among them, G is the characteristic number of genes, μ s and μ t is the average value of β,

[0051] Finally, a q×q dimensional matrix will be obtained, where q represents the characteristic dimension of the brain image.

[0052] Gene-gene association matrix: XX represents the gene-gene association matrix. Reference gene data downloaded from the 1000 Genomes Project were integrated into a p×p-dimensional matrix according to the gene sequences in the gene-brain image association matrix. Here, p represents the characteristic dimension of the gene.

[0053] In an optional embodiment, before generating a gene-brain imaging association matrix based on the weight values ​​corresponding to the same-modality GWAS data, the method further includes: determining whether the gene data and brain imaging data were normalized before the GWAS was performed; if the gene data and brain imaging data were not normalized, then correcting the weight values ​​corresponding to the GWAS data. Furthermore, the weight values ​​corresponding to the GWAS data can be corrected using the following correction formula:

[0054]

[0055] Among them, β STANDR represents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

[0056] It should be noted that before using the β value, it is necessary to check whether the genes and brain images are normalized before the GWAS analysis. If normalization is not performed, the β value needs to be corrected.

[0057] In an optional embodiment, a multimodal brain imaging-gene association model is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix based on the co-expressed gene set to obtain the target brain region and the gene loci of the target brain region, including: determining the objective function of the multimodal brain imaging-gene association model, wherein the expression of the objective function is as follows:

[0058]

[0059] in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the characteristic dimension of the gene, q represents the characteristic dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modalities, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges to obtain the target brain region and the gene loci of the target brain region.

[0060] In the specific implementation process, the multimodal brain imaging-gene association method of the present invention is used to perform association analysis on the collated data. The specific implementation algorithm formula is:

[0061]

[0062]

[0063] Writing this objective function in augmented Lagrangian form, the objective function becomes:

[0064]

[0065] It should be noted that in the above embodiment, the original gene features and brain image features are not directly input, but the gene-brain image correlation matrix, brain image-brain image correlation matrix, and gene-gene correlation matrix are input.

[0066] In addition, λ is used to constrain brain imaging features, and γ is used to constrain genotype features.

[0067] In an optional embodiment, the gene-brain image correlation matrix, the brain image-brain image correlation matrix and the gene-gene correlation matrix are respectively input into the objective function for processing, including: inputting λ1, ..., λ m ,γ,α,τ1,...,τ m,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialize u∈R p×1 and v1,...,v m ∈R q×1 ; At this point the loop begins, and the end condition is that the objective function converges; update D G , where D G is a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; solve for u so that Update D m , where D m is a diagonal matrix, each diagonal element is Solve for v1,...,v m , and make

[0068] It should be noted that

[0069] An optional implementation manner of the present invention is described in detail below.

[0070] Figure 2 FIG. 1 is a flow chart of another method for processing multimodal brain imaging and gene data according to an embodiment of the present invention. Figure 2 As shown, the specific implementation steps of this method are as follows:

[0071] Step 1: Data Collection and Organization. Collect GWAS analysis data results related to multimodal brain imaging. Based on the gene sequences in the collected GWAS data, download the corresponding genetic data of the reference population.

[0072] Step 2: Use the collected GWAS analysis data results and the genetic data of the corresponding reference population to organize them into brain image-gene association matrix, brain image-brain image association matrix, and gene-gene association matrix.

[0073] Step 3: Based on the collected gene data sequences, pathway metabolic pathway analysis is performed to obtain gene enrichment pathways, and then the co-expressed genes are grouped using the known gene enrichment pathway results.

[0074] Step 4: Analyze the above data using the multimodal brain imaging-gene association analysis model proposed in this invention. After the model converges, the diseased brain regions and gene loci associated with the disease are ultimately derived.

[0075] The present invention adopts the above technical solution to have the following advantages and beneficial effects:

[0076] (1) The present invention combines a multimodal approach to fully utilize the advantages of multiple modalities and fuses them for prediction, which can achieve good prediction results;

[0077] (2) When considering the effect of genes on diseases, the present invention not only considers the effect of a single gene on the disease, but also further considers the effect of multiple genes on the disease. Therefore, the gene influence considered by the present invention is more comprehensive.

[0078] (3) Current research on multimodal brain imaging and genes is based on association analysis using raw genetic and brain imaging data. However, the present invention uses existing genome-wide association study (GWAS) data for analysis, eliminating the need for raw data and leveraging existing publicly available datasets. Therefore, the present invention has a wider scope of application.

[0079] According to another aspect of the embodiments of the present invention, a device for processing multimodal brain images and genetic data is provided. Figure 3 is a schematic diagram of a multimodal brain image and gene data processing device according to an embodiment of the present invention, such as Figure 3 As shown, the multimodal brain image and gene data processing device includes: an acquisition module 32, a creation module 34, a grouping module 36, and a processing module 38. The multimodal brain image and gene data processing device is described in detail below.

[0080] An acquisition module 32 is used to obtain genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population; an establishment module 34 is connected to the acquisition module 32 and is used to establish a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix and a gene-gene association matrix based on the GWAS data and the genetic data; a grouping module 36 is connected to the establishment module 34 and is used to perform pathway metabolic pathway analysis on the genetic data, and group the co-expressed genes according to the analysis results to obtain a co-expressed gene set; a processing module 38 is connected to the grouping module 36 and is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix and the gene-gene association matrix based on the co-expressed gene set using a multimodal brain imaging-gene association model to obtain target brain regions and gene loci in the target brain regions.

[0081] It should be noted that the above-mentioned modules can be implemented by software or hardware. For example, for the latter, it can be implemented in the following ways: the above-mentioned modules can be located in the same processor; and / or the above-mentioned modules can be located in different processors in any combination.

[0082] In the above embodiment, the multimodal brain image and gene data processing device can obtain genome-wide association analysis GWAS data of multimodal brain images and gene data of corresponding reference populations; establish a gene-brain image association matrix, a brain image-brain image association matrix, and a gene-gene association matrix based on the GWAS data and gene data; perform pathway metabolic pathway analysis on the gene data, and group the co-expressed genes according to the analysis results to obtain a co-expressed gene set; based on the co-expressed gene set, use a multimodal brain image-gene association model to process the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix to obtain the target brain region and the target brain region's gene loci. In other words, the embodiment of the present invention uses metabolic pathways to divide the co-expressed gene set and conducts association analysis based on multimodal brain images. Specifically, through the collected and organized data, the model is used to automatically derive the target brain region and the target brain region's gene loci, thereby solving the problem of difficulty in obtaining data and poor processing effect in the related art of association analysis based on raw gene data and single-modality brain image data, thereby achieving the technical effect of reducing the difficulty of data acquisition and improving the accuracy of prediction results.

[0083] It should be noted here that the above-mentioned acquisition module 32, establishment module 34, grouping module 36 and processing module 38 correspond to steps S102 to S108 in the method embodiment. The examples and application scenarios implemented by the above-mentioned modules and corresponding steps are the same, but are not limited to the contents disclosed in the above-mentioned method embodiment.

[0084] Optionally, the acquisition module 32 includes: a first acquisition unit for acquiring GWAS data of multimodal brain imaging; and a second acquisition unit for acquiring gene data of a corresponding reference population based on gene sequences in the GWAS data.

[0085] Optionally, the above-mentioned establishment module 34 includes: a first establishment unit, used to generate a gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; a second establishment unit, used to transform the gene-brain image association matrix to obtain a brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; a third establishment unit, used to generate a gene-gene association matrix based on the gene data and the gene sequence in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p-dimensional matrix; wherein p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain image.

[0086] Optionally, the above-mentioned establishment module 34 also includes: a judgment unit, which is used to judge whether the gene data and brain imaging data are normalized before performing GWAS before generating the gene-brain imaging association matrix based on the weight values ​​corresponding to the GWAS data of the same modality; and a correction unit, which is used to correct the weight values ​​corresponding to the GWAS data if the gene data and brain imaging data are not normalized.

[0087] Optionally, the weight values ​​corresponding to the GWAS data are corrected using the following correction formula: Among them, β STANDR represents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

[0088] Optionally, the processing module 38 includes: a determination unit, configured to determine an objective function of the multimodal brain image-gene association model, wherein the objective function is expressed as follows:

[0089]

[0090] in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the characteristic dimension of the gene, q represents the characteristic dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modes, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; the processing unit is used to input the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix into the objective function for processing, until the objective function converges, and the target brain region and the gene loci of the target brain region are obtained.

[0091] Optionally, the processing unit includes an input subunit for inputting λ1, ..., λ into the objective function respectively. m ,γ,α,τ1,...,τ m ,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialization subunit, used to initialize u∈R p×1 and v1,...,v m ∈R q ×1 ; The first update subunit is used to update D G , where D Gis a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; the first solving subunit is used to solve u and make The second updating subunit is used to update D m , where D m is a diagonal matrix, each diagonal element is The second solving subunit is used to solve v1,...,v m , and make

[0092] According to another aspect of an embodiment of the present invention, an electronic device is also provided, which includes: a processor; a memory for storing instructions executable by the processor; wherein the processor is configured to execute any one of the above-mentioned methods for processing multimodal brain imaging and genetic data.

[0093] Optionally, an embodiment of the present invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and runnable on the processor, and when the processor executes the program, the following steps are implemented: obtaining genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population; establishing a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix, and a gene-gene association matrix based on the GWAS data and the genetic data; performing pathway metabolic pathway analysis on the genetic data, and grouping the co-expressed genes according to the analysis results to obtain a co-expressed gene set; based on the co-expressed gene set, using a multimodal brain imaging-gene association model to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix to obtain target brain regions and gene loci of the target brain regions.

[0094] Optionally, obtaining genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population includes: obtaining GWAS data of multimodal brain imaging; and obtaining genetic data of a corresponding reference population based on gene sequences in the GWAS data.

[0095] Optionally, based on the GWAS data and the genetic data, a gene-brain image association matrix, a brain image-brain image association matrix and a gene-gene association matrix are established respectively, including: generating a gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; transforming the gene-brain image association matrix to obtain a brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; generating a gene-gene association matrix based on the genetic data and the gene sequence in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p-dimensional matrix; wherein p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain image.

[0096] Optionally, before generating a gene-brain imaging association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, the above method also includes: determining whether the gene data and brain imaging data are normalized before performing GWAS; if the gene data and brain imaging data are not normalized, correcting the weight values ​​corresponding to the GWAS data.

[0097] Optionally, the weight values ​​corresponding to the GWAS data are corrected using the following correction formula: Among them, β STANDR represents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

[0098] Optionally, based on the co-expressed gene set, a multimodal brain imaging-gene association model is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix to obtain the target brain region and the gene loci of the target brain region, including: determining the objective function of the multimodal brain imaging-gene association model, wherein the expression of the objective function is as follows:

[0099]

[0100] in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the characteristic dimension of the gene, q represents the characteristic dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modalities, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges to obtain the target brain region and the gene loci of the target brain region.

[0101] Optionally, the gene-brain image correlation matrix, the brain image-brain image correlation matrix, and the gene-gene correlation matrix are respectively input into the objective function for processing, including: inputting λ1, ..., λ into the objective function respectively. m ,γ,α,τ1,...,τ m ,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialize u∈R p×1 and v1,...,v m ∈R q×1 ; Update D G , where D G is a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; solve for u so that Update D m , where D m is a diagonal matrix, each diagonal element is Solve for v1,...,v m , and make

[0102] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is also provided, which includes a stored program, wherein when the program is running, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned methods for processing multimodal brain images and genetic data.

[0103] Optionally, in this embodiment, the computer-readable storage medium may be located in any one of the computer terminals in the computer terminal group in the computer network, and / or located in any one of the mobile terminals in the mobile terminal group, and the computer-readable storage medium includes a stored program.

[0104] Optionally, when the program is running, the device where the computer-readable storage medium is located is controlled to perform the following functions: obtain genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of the corresponding reference population; establish a gene-brain imaging association matrix, a brain imaging-brain imaging association matrix and a gene-gene association matrix based on the GWAS data and the genetic data, respectively; perform pathway metabolic pathway analysis on the genetic data, and group the co-expressed genes according to the analysis results to obtain a co-expressed gene set; based on the co-expressed gene set, use a multimodal brain imaging-gene association model to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix and the gene-gene association matrix to obtain the target brain region and the gene loci of the target brain region.

[0105] Optionally, obtaining genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of a corresponding reference population includes: obtaining GWAS data of multimodal brain imaging; and obtaining genetic data of a corresponding reference population based on gene sequences in the GWAS data.

[0106] Optionally, based on the GWAS data and the genetic data, a gene-brain image association matrix, a brain image-brain image association matrix and a gene-gene association matrix are established respectively, including: generating a gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; transforming the gene-brain image association matrix to obtain a brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; generating a gene-gene association matrix based on the genetic data and the gene sequence in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p-dimensional matrix; wherein p represents the characteristic dimension of the gene, and q represents the characteristic dimension of the brain image.

[0107] Optionally, before generating a gene-brain imaging association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, the above method also includes: determining whether the gene data and brain imaging data are normalized before performing GWAS; if the gene data and brain imaging data are not normalized, correcting the weight values ​​corresponding to the GWAS data.

[0108] Optionally, the weight values ​​corresponding to the GWAS data are corrected using the following correction formula: Among them, β STANDR represents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

[0109] Optionally, based on the co-expressed gene set, a multimodal brain imaging-gene association model is used to process the gene-brain imaging association matrix, the brain imaging-brain imaging association matrix, and the gene-gene association matrix to obtain the target brain region and the gene loci of the target brain region, including: determining the objective function of the multimodal brain imaging-gene association model, wherein the expression of the objective function is as follows:

[0110]

[0111] in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the characteristic dimension of the gene, q represents the characteristic dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modalities, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges to obtain the target brain region and the gene loci of the target brain region.

[0112] Optionally, the gene-brain image correlation matrix, the brain image-brain image correlation matrix, and the gene-gene correlation matrix are respectively input into the objective function for processing, including: inputting λ1, ..., λ into the objective function respectively. m ,γ,α,τ1,...,τ m ,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialize u∈R p×1 and v1,...,v m ∈R q×1 ; Update D G , where D G is a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; solve for u so that Update D m , where D m is a diagonal matrix, each diagonal element is Solve for v1,...,v m , and make

[0113] The above description is merely a preferred embodiment of the present invention and is not intended to limit the scope of protection of the present invention.

Claims

1. A method for processing multimodal brain imaging and genetic data, characterized in that: include: Obtain genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of corresponding reference populations; Based on the GWAS data and the gene data, a gene-brain image association matrix, a brain image-brain image association matrix, and a gene-gene association matrix are respectively established; including: Generating the gene-brain image association matrix according to the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; Performing a transformation process on the gene-brain image association matrix to obtain the brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; Generating a gene-gene association matrix based on the gene data and the gene sequences in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p dimensional matrix; Among them, p represents the characteristic dimension of genes, and q represents the characteristic dimension of brain images; Performing pathway metabolic pathway analysis on the gene data, and grouping the co-expressed genes according to the analysis results to obtain a co-expressed gene set; The gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are processed using a multimodal brain image-gene association model based on the co-expressed gene set to obtain a target brain region and a gene locus of the target brain region; comprising: Determine the objective function of the multimodal brain imaging-gene association model, wherein the objective function is expressed as follows: in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the feature dimension of the gene, q represents the feature dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modalities, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; The gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges, thereby obtaining the target brain region and the gene loci of the target brain region.

2. The method according to claim 1, characterized in that Obtain genome-wide association analysis GWAS data for multimodal brain imaging and genetic data of corresponding reference populations, including: obtaining the GWAS data of the multimodal brain imaging; According to the gene sequence in the GWAS data, the gene data corresponding to the reference population is obtained.

3. The method according to claim 2, characterized in that Before generating the gene-brain image association matrix based on the weight values ​​corresponding to the GWAS data of the same modality, the method further includes: Determining whether to normalize the genetic data and the brain imaging data before performing GWAS; If the gene data and the brain imaging data are not normalized, the weight values ​​corresponding to the GWAS data are corrected.

4. The method according to claim 3, characterized in that The weight values ​​corresponding to the GWAS data were corrected using the following correction formula: Among them, β STANDR represents the weight value corresponding to the corrected GWAS data, β represents the weight value corresponding to the current GWAS data, N represents the number of samples, SE β represents the standard error of β.

5. The method according to claim 1, characterized in that Inputting the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix into the objective function for processing respectively includes: Input λ1,...,λ to the objective function respectively m ,γ,α,τ1,...,τ m ,XX,XY1,...,XY m ,Y1Y1,...,Y m Y m ; Initialize u∈R p×1 and v1,...,v m ∈R q×1 ; Update D G , where D G is a diagonal matrix, and the g-th diagonal block is I g is the identity matrix of the g-th diagonal block; Solve for u so that Update D m , where D m is a diagonal matrix, each diagonal element is Solve for v1,...,v m , and make 6. A multimodal brain image and gene data processing device, characterized in that: include: The acquisition module is used to obtain genome-wide association analysis GWAS data of multimodal brain imaging and genetic data of the corresponding reference population; An establishment module is used to establish a gene-brain image association matrix, a brain image-brain image association matrix, and a gene-gene association matrix based on the GWAS data and the gene data; including: Generating the gene-brain image association matrix according to the weight values ​​corresponding to the GWAS data of the same modality, wherein the gene-brain image association matrix is ​​a p×q-dimensional matrix; Performing a transformation process on the gene-brain image association matrix to obtain the brain image-brain image association matrix, wherein the brain image-brain image association matrix is ​​a q×q-dimensional matrix; Generating a gene-gene association matrix based on the gene data and the gene sequences in the gene-brain image association matrix, wherein the gene-gene association matrix is ​​a p×p dimensional matrix; Among them, p represents the characteristic dimension of genes, and q represents the characteristic dimension of brain images; A grouping module is used to perform pathway metabolic pathway analysis on the gene data, and group the co-expressed genes according to the analysis results to obtain a co-expressed gene set; A processing module is configured to process the gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix using a multimodal brain image-gene association model based on the co-expressed gene set to obtain a target brain region and a gene locus of the target brain region; comprising: Determine the objective function of the multimodal brain imaging-gene association model, wherein the objective function is expressed as follows: in, represents the gene feature matrix, represents the brain image feature matrix, n represents the number of samples corresponding to the co-expressed gene set, p represents the feature dimension of the gene, q represents the feature dimension of the brain image, v represents the coefficient vector corresponding to the brain image feature, u represents the coefficient vector corresponding to the gene feature, m represents the number of brain image modalities, λ represents the first regularization parameter, γ represents the second regularization parameter, α represents the third regularization parameter, and τ represents the fourth regularization parameter; The gene-brain image association matrix, the brain image-brain image association matrix, and the gene-gene association matrix are respectively input into the objective function for processing until the objective function converges, thereby obtaining the target brain region and the gene loci of the target brain region.

7. An electronic device, characterized in that: include: processor; a memory for storing processor-executable instructions; Wherein, the processor is configured to execute the multimodal brain imaging and genetic data processing method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that The computer-readable storage medium includes a stored program, wherein when the program is executed, the device where the computer-readable storage medium is located is controlled to execute the method for processing multimodal brain imaging and genetic data according to any one of claims 1 to 5.

Images