A gene co-expression network identification method and system based on graph convolutional neural network

Abstract

The present application relates to a kind of gene co-expression network identification method and system based on graph convolutional neural network, the method includes steps: S1.data preprocessing;S2.based on gene expression correlation constructs gene network diagram;S3.captures node information by graph convolutional neural network GraphSAGE model and extracts feature;S4.cluster analysis is carried out using Gaussian mixture model;S5.according to clustering result, gene module is divided.This method can help researchers more effectively from mass gene expression data extract valuable information, in-depth understanding of gene regulatory network, discover new gene function and regulatory mechanism, provide important guidance for the treatment and prevention of disease.Also have good scalability and flexibility, can adapt to different scale and type of gene expression data analysis needs.

Description

Technical Field

[0001] This invention belongs to the field of bioinformatics, specifically relating to a method and system for identifying gene co-expression networks based on graph convolutional neural networks. Background Technology

[0002] In recent years, with the rapid development of bioinformatics, the analysis of gene expression data has become one of the important means to unravel the mysteries of life sciences. Gene expression data reflects the expression level of genes in an organism under different conditions, and is of great significance for understanding gene function, regulatory mechanisms, and disease pathogenesis. However, given the massive complexity of gene expression data, extracting valuable information from such vast amounts of data remains one of the challenges facing the field of bioinformatics. Currently, widely used gene co-expression network analysis methods are mainly based on biostatistical techniques, which face problems such as high computational complexity and insufficient accuracy when processing large-scale genomic data. Summary of the Invention

[0003] To address the shortcomings of existing technologies, this invention introduces GraphSAGE, a graph convolutional network from the field of deep learning, to perform neighbor sampling and aggregation for feature selection. Based on this, it identifies gene co-expression modules, which is expected to reveal the regulatory relationships between genes more accurately and efficiently, thereby discovering some previously unknown gene functions, and providing important guidance for the treatment and prevention of diseases.

[0004] The technical solution of this invention is as follows:

[0005] A gene co-expression network identification method based on graph convolutional neural networks includes the following steps:

[0006] S1. Data preprocessing to reduce data noise and scale differences.

[0007] S2. Constructing a gene network graph based on gene expression correlations: Based on preprocessed gene expression data, a gene network graph is constructed by calculating the correlations between genes (such as the Pearson correlation coefficient). In the graph, genes are treated as nodes, and the correlations between genes are used as the weights of the edges.

[0008] S3. Capturing node information and extracting features using the GraphSAGE convolutional neural network model; The GraphSAGE model is used to capture and extract features from node information in the gene network graph, combined with a sampler to sample and aggregate the neighbors of nodes. GraphSAGE can learn the features of nodes and their neighboring nodes, thereby obtaining the complex regulatory relationships between genes.

[0009] S4. Cluster analysis using Gaussian Mixture Model (GMM); GMM is used to perform cluster analysis on the features extracted by the graph convolutional network. GMM can fit the complex distribution of the data and identify different gene modules. The clustering results divide the genes into different co-expression modules.

[0010] S5. Divide the gene modules according to the clustering results; perform principal component analysis on each cluster module to obtain the first principal component, then calculate the correlation coefficient between the sample label and the first principal component, and finally take the module genes with high correlation as the markers of the gene network, which can be used for subsequent downstream analysis.

[0011] According to an embodiment of the present invention, in step S1, the preprocessing includes data cleaning, missing value imputation, standardization, and variance screening. Preferably, it includes first performing mean screening on the data, and then performing variance screening to rank the genes by variance and screen out genes with high variance to ensure the accuracy and comparability of the data. In a preferred embodiment, the homogeneity screening threshold is 0.5 for the fpkm mean, and the top 10,000 genes with high variance are selected.

[0012] According to an embodiment of the present invention, in step S2, using genes as nodes, the Pearson correlation coefficient C is calculated for all gene pairs. Then, using a given soft threshold st, the P value is calculated based on this correlation coefficient to determine the edge. If the P value is greater than the hard threshold ht, the edge is recorded in the gene network graph, and the P value is used as the edge weight. The P value is calculated using the following formula: P = |Cij|^st. In a preferred embodiment, the given soft threshold st = 7, and the hard threshold ht is set to 0.01.

[0013] According to an embodiment of the present invention, in step S3, the GraphSAGE model includes an input layer, two convolutional layers SAGEConv, and an output layer. The input layer receives a node feature matrix x. Each convolutional layer SAGEConv is followed by a network layer for normalizing the features, increasing model nonlinearity, and preventing overfitting. Finally, the output layer normalizes the feature dimension to obtain the final graph embedding, which is represented as low-dimensional node features. In a preferred embodiment, both convolutional layers sample 10 neighbors each, sampling only 128 nodes at a time. The output feature dimension after the two convolutional layers SAGEConv is 128; the output layer normalizes the feature dimension to 5.

[0014] According to an embodiment of the present invention, in step S4, the features extracted by the graph convolutional network are clustered using a Gaussian mixture model, dividing the genes into different co-expression modules. The number of clusters, k, is set to twice the square root of the number of graph embeddings, and the covariance type is set to diagonal to allow for uncorrelated dimensions within each cluster. The prior weight concentration 1 / k is adjusted based on the maximum number of clusters to control the weight distribution among different clusters. Furthermore, soft clustering is used to assign a probability distribution belonging to each cluster to each data point.

[0015] According to an embodiment of the present invention, in step S5, the significance score of the clustering module is calculated. Principal component analysis is performed on each clustering module to obtain the first principal component. Then, the Pearson correlation coefficient is calculated using the sample label and the first principal component. Finally, the module genes with high correlation are taken as markers of the gene network, which can be used for subsequent downstream analysis.

[0016] In another aspect, the present invention provides a gene co-expression network identification system based on graph convolutional neural networks, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements the above-described gene co-expression network identification method based on graph convolutional neural networks.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] The analytical method provided by this invention enables researchers to more effectively extract valuable information from massive gene expression data, gain a deeper understanding of gene regulatory networks, discover new gene functions and regulatory mechanisms, and provide important guidance for disease treatment and prevention. Furthermore, the gene co-expression network identification method of this invention also possesses good scalability and flexibility, adapting to the needs of gene expression data analysis of different scales and types. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the technical process of an embodiment of the present invention. Detailed Implementation

[0020] To facilitate understanding of the present invention, a more complete description will be given below with reference to the accompanying drawings. Preferred embodiments of the invention are shown in the drawings. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a thorough and complete understanding of the disclosure of the invention.

[0021] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains. The terminology used herein in the description of the invention is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.

[0022] Unless otherwise specified, all reagents, materials, and equipment used in the embodiments of this invention are commercially available; unless otherwise specified, all test methods are conventional test methods in the art.

[0023] Reference Figure 1 This invention provides a gene co-expression network identification method based on graph convolutional neural networks, comprising the following steps:

[0024] S1. Data preprocessing.

[0025] Preprocessing can include data cleaning, missing value imputation, standardization, and variance filtering, which can be selected as needed to reduce data noise and scale differences.

[0026] Before performing variance filtering, the data should first be filtered for mean to ensure the accuracy and comparability of the data.

[0027] In one embodiment, the data is first filtered by mean with a threshold of 0.5 for the fpkm mean, and then by variance filtering. The genes are sorted by variance, and 10,000 genes with high variance are selected to ensure the accuracy and comparability of the data.

[0028] S2. Construct a gene co-expression network.

[0029] Based on preprocessed gene expression data, a gene network graph is constructed by calculating the correlations between genes (such as the Pearson correlation coefficient). In the graph, genes are treated as nodes, and the correlations between genes are used as the weights of the edges.

[0030] Specifically, using genes as nodes, the Pearson correlation coefficient C is calculated for all gene pairs. Then, a P value is calculated based on this correlation coefficient using a given soft threshold st to determine the edge. If the P value is greater than the hard threshold ht, the edge is recorded in the gene network graph, and the P value is used as the edge weight. The P value is calculated using the following formula: P = |Cij|^st. In a preferred embodiment, the given soft threshold st = 7, and the hard threshold ht is set to 0.01.

[0031] S3. Low-dimensional feature representation

[0032] The GraphSAGE model, a graph convolutional neural network, is used to capture and extract features from nodes in a gene network graph. A sampler is then used to sample and aggregate the neighbors of each node. GraphSAGE can learn the features of nodes and their neighbors, thereby revealing the complex regulatory relationships between genes.

[0033] In one embodiment, the GraphSAGE model includes an input layer, two convolutional layers (SAGEConv), and an output layer. The input layer receives a node feature matrix x. Each SAGEConv convolutional layer is followed by a network layer for normalizing the features, increasing model nonlinearity, and preventing overfitting. Finally, the output layer normalizes the feature dimension to obtain the final graph embedding, which is represented as low-dimensional node features. In one embodiment, both convolutional layers sample 10 neighbors each, sampling only 128 nodes at a time. The output feature dimension after the two SAGEConv convolutional layers is 128; the output layer normalizes the feature dimension to 5.

[0034] S4. Use Gaussian mixture model for cluster analysis.

[0035] Cluster analysis was performed on the features extracted by the graph convolutional network using a Gaussian mixture model (GMM). GMM is able to fit the complex distribution of the data and identify different gene modules. The clustering results divided the genes into different co-expression modules.

[0036] In one embodiment, the number of clusters k is set to twice the square root of the number of graph embeddings, and the covariance type is set to diagonal to allow for uncorrelated dimensions within each cluster. The prior weight concentration 1 / k is adjusted based on the maximum number of clusters to control the weight distribution among different clusters. Furthermore, soft clustering is used to assign a probability distribution belonging to each cluster to each data point.

[0037] S5. Divide the gene modules according to the clustering results.

[0038] Principal component analysis is performed on each cluster module to obtain the first principal component. Then, the correlation coefficient, such as the Pearson correlation coefficient, is calculated between the sample labels and the first principal component. Finally, the genes of the modules with high correlation are selected as markers of the gene network, which can be used for subsequent downstream analysis.

[0039] The present invention also provides a gene co-expression network identification system based on graph convolutional neural networks, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the computer program is loaded onto the processor, it implements the above-described gene co-expression network identification method based on graph convolutional neural networks.

[0040] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0041] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these all fall within the protection scope of the present invention. Therefore, the protection scope of this invention patent should be determined by the appended claims.

Claims

1. A gene co-expression network identification method based on graph convolutional neural networks, characterized in that, Includes the following steps: S1. Data preprocessing; S2. Construct a gene network graph based on gene expression correlation; using genes as nodes, calculate the Pearson correlation coefficient C for all gene pairs, and then use the given soft threshold st to calculate the P value of the correlation coefficient C to determine the edge. If the P value is greater than the hard threshold ht, the edge is recorded in the gene network graph and the P value is used as the weight of the edge. The formula for calculating the P value is as follows: P = |Cij|^st. S3. Capture node information and extract features using the GraphSAGE convolutional neural network model; The GraphSAGE model includes an input layer, two convolutional layers SAGEConv, and an output layer; The input layer accepts the node feature matrix x, and each convolutional layer SAGEConv is followed by a network layer for normalizing the features, increasing the model's non-linearity, and preventing overfitting. Finally, the output layer normalizes the feature dimensions to obtain the final graph embedding, which is represented as low-dimensional node features; S4. Use Gaussian mixture model for cluster analysis; set the number of clusters k to twice the square root of the number of graph embeddings, set the covariance type to diagonal, adjust the prior weight concentration 1 / k according to the maximum number of clusters, and use soft clustering to assign a probability distribution to each data point belonging to each cluster. S5. Divide the gene modules according to the clustering results; perform principal component analysis on each cluster module to obtain the first principal component, then calculate the Pearson correlation coefficient with the first principal component using the sample label, and finally take the module genes with high correlation as the markers of the gene network for subsequent downstream analysis.

2. The gene co-expression network identification method according to claim 1, characterized in that, In step S1, preprocessing includes data cleaning, missing value imputation, standardization, and variance filtering.

3. The gene co-expression network identification method according to claim 2, characterized in that, First, the data were homogeneously screened, and then variance-screened. The homogeneity screening threshold was 0.5 for the mean of fpkm, and the variance screening identified 10,000 genes with high variance.

4. The gene co-expression network identification method according to claim 1, characterized in that, The given soft threshold st=7, and the hard threshold ht is set to 0.

01.

5. The gene co-expression network identification method according to claim 1, characterized in that, Both convolutional layers sample 10 neighbors each, sampling only 128 nodes at a time. After two convolutional layers SAGEConv, the output feature dimension is 128; the output layer normalizes the feature dimension to 5.

6. A gene co-expression network recognition system based on graph convolutional neural networks, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the computer program is loaded into the processor, it implements the gene co-expression network identification method based on graph convolutional neural networks according to any one of claims 1-5.

Images