Tissue gene transcription factor identification method and device

Abstract

The invention discloses a method and a device for identifying a transcription factor of a tissue gene, relates to the technical field of biological information, and mainly aims to solve the problem of poor identification accuracy of the transcription factor of the existing tissue gene. Comprising the following steps: acquiring an initial gene expression matrix of tissue genes, and constructing a full-connection directed graph based on the initial gene expression matrix; after the full-connection directed graph is converted into an undirected role graph, the undirected role graph and a feature matrix are reconstructed based on a graph neural network model after model training is completed, a reconstructed target gene expression matrix is obtained, and a loss value of the graph neural network model is obtained through calculation after edges are dynamically removed; and generating a directed gene regulatory network based on the target gene expression matrix, and identifying a transcription factor of the tissue gene based on the directed gene regulatory network.

Description

Technical Field

[0001] The present application relates to the field of bioinformatics, and in particular to a method and device for identifying transcription factors of tissue genes. Background Art

[0002] With the development of single-cell RNA sequencing (scRNA-seq) technology, gene regulatory network (GRN) inference has become a key means to gain a deeper understanding of cell fate determination, disease mechanisms, and the regulation of biological processes. Currently, in order to accurately infer the regulatory relationships of tissue genes, deep learning can usually be used to identify transcription factors using prior knowledge or high-quality annotated data. However, prior knowledge cannot meet the requirements for applications in new cell types or diseases, and the high sparsity of single-cell data makes deep learning algorithms susceptible to noise, making it difficult to distinguish between direct and indirect regulatory relationships, greatly reducing the accuracy of transcription factor identification and making it unable to adapt to the dynamic identification needs of biological tissue gene regulation. Therefore, a method for identifying transcription factors of tissue genes is urgently needed to solve the above technical problems. Summary of the Invention

[0003] In view of this, the present application provides a method and device for identifying transcription factors of tissue genes, the main purpose of which is to solve the problem of poor accuracy in the existing identification of transcription factors of tissue genes.

[0004] According to one aspect of the present application, a method for identifying transcription factors of tissue genes is provided, comprising: Obtaining an initial gene expression matrix of tissue genes, and constructing a fully connected directed graph based on the initial gene expression matrix; After converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph and the feature matrix based on the graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; A directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network.

[0005] Furthermore, after converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph based on the graph neural network model for which model training has been completed to obtain a reconstructed target gene expression matrix, the method further includes: Acquire a sample data set, wherein the sample data set includes an undirected role graph sample and a feature matrix sample; Construct a two-layer graph neural network, where the first layer of the two-layer graph neural network is used to aggregate neighbor node features, and the second layer of the two-layer graph neural network is used to generate high-order feature representations; During the model training of the two-layer graph neural network based on the sample data set, all network edges are traversed, and when the target network edge is deleted to determine that the loss value is minimum, the model training of the two-layer graph neural network is completed to obtain a graph neural network model that has completed model training.

[0006] Furthermore, the traversal of all network edges and the completion of model training of the two-layer graph neural network when the target network edge is deleted to determine that the loss value is the minimum, and the obtained graph neural network model that completes model training includes: Determine the initial loss baseline value; Deleting a target network edge, and calculating a loss value of the two-layer graph neural network corresponding to deleting the target network edge; When the loss value is less than the initial loss reference value, the initial loss reference value is updated based on the loss value, and the network edge is deleted; When the loss value is greater than or equal to the initial loss reference value, the target network edge is updated, and the step of calculating and deleting the loss value of the two-layer graph neural network corresponding to the target network edge is re-executed.

[0007] Furthermore, after constructing the fully connected directed graph based on the initial gene expression matrix, the method further includes: Splitting the nodes in the fully connected directed graph to obtain input nodes and output nodes; The input node and the output node are encoded based on a bidirectional encoder to obtain the undirected role graph and the feature matrix.

[0008] Furthermore, before obtaining the initial gene expression matrix of tissue genes, the method further includes: Determine RNA sequence data of tissue genes; Deleting the RNA sequence data of low-expressed genes and abnormal cell genes from the RNA sequence data based on a preset pipeline, and normalizing and standardizing the deleted RNA sequence data to obtain gene data; The initial gene expression matrix is ​​constructed based on the gene data.

[0009] Furthermore, constructing a fully connected directed graph based on the initial gene expression matrix includes: The network edges in the fully connected directed graph are screened based on the Pearson correlation coefficient or the protein interaction coefficient to obtain a fully connected directed graph to be converted into an undirected graph.

[0010] Furthermore, the identifying the transcription factor of the tissue gene based on the directed gene regulatory network includes: determining the out-degree of the tissue gene according to the directed gene regulatory network; By comparing the out-degrees horizontally, the maximum out-degree in the directed gene regulatory network is determined, and the transcription factor of the tissue gene is determined based on the cell type corresponding to the maximum out-degree.

[0011] According to another aspect of the present application, a device for identifying transcription factors of tissue genes is provided, comprising: an acquisition module, configured to acquire an initial gene expression matrix of tissue genes and construct a fully connected directed graph based on the initial gene expression matrix; A reconstruction module is used to reconstruct the undirected role graph and the feature matrix based on the graph neural network model that has completed model training after converting the fully connected directed graph into an undirected role graph to obtain a reconstructed target gene expression matrix, wherein the loss value of the graph neural network model is calculated after dynamically removing edges; An identification module is used to generate a directed gene regulatory network based on the target gene expression matrix, and identify the transcription factors of the tissue genes based on the directed gene regulatory network.

[0012] Furthermore, the device further comprises: a construction module, a training module, The acquisition module is further configured to acquire a sample data set, wherein the sample data set includes an undirected role graph sample and a feature matrix sample; The building module is used to construct a two-layer graph neural network, wherein the first layer of the two-layer graph neural network is used to aggregate neighbor node features, and the second layer of the two-layer graph neural network is used to generate high-order feature representations; The training module is used to traverse all network edges during the model training of the two-layer graph neural network based on the sample data set, and complete the model training of the two-layer graph neural network when the target network edge is deleted to determine that the loss value is minimized, so as to obtain a graph neural network model that has completed model training.

[0013] Furthermore, the training module is specifically used to determine an initial loss benchmark value; delete the target network edge, and calculate the loss value of the two-layer graph neural network corresponding to the deletion of the target network edge; when the loss value is less than the initial loss benchmark value, the initial loss benchmark value is updated based on the loss value, and the network edge is deleted; when the loss value is greater than or equal to the initial loss benchmark value, the target network edge is updated, and the step of calculating the loss value of the two-layer graph neural network corresponding to the deletion of the target network edge is re-executed.

[0014] Furthermore, the device further comprises: A splitting module, configured to split the nodes in the fully connected directed graph to obtain input nodes and output nodes; An encoding module is used to encode the input node and the output node based on a bidirectional encoder to obtain the undirected role graph and the feature matrix.

[0015] Furthermore, the device further includes: a determination module, a processing module, The determination module is used to determine the RNA sequence data of the tissue gene; The processing module is used to delete the RNA sequence data of low-expressed genes and abnormal cell genes in the RNA sequence data based on a preset pipeline, and normalize and standardize the deleted RNA sequence data to obtain gene data; The construction module is further used to construct the initial gene expression matrix based on the gene data.

[0016] Furthermore, the construction module is also used to screen the network edges in the fully connected directed graph based on the Pearson correlation coefficient or the protein interaction coefficient to obtain a fully connected directed graph to be converted into an undirected graph.

[0017] Furthermore, the identification module is specifically used to determine the out-degree of the tissue gene based on the directed gene regulatory network; determine the maximum out-degree in the directed gene regulatory network by horizontally comparing the out-degrees, and determine the transcription factor of the tissue gene based on the cell type corresponding to the maximum out-degree.

[0018] According to another aspect of the present application, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the above-mentioned method for identifying transcription factors of tissue genes.

[0019] According to another aspect of the present application, there is provided a terminal, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned method for identifying transcription factors of tissue genes.

[0020] By means of the above technical solution, the technical solution provided by the embodiment of the present application has at least the following advantages: The present application provides a method and device for identifying transcription factors of tissue genes. Compared with the prior art, the embodiment of the present application obtains the initial gene expression matrix of the tissue genes and constructs a fully connected directed graph based on the initial gene expression matrix; after converting the fully connected directed graph into an undirected role graph, the undirected role graph and the feature matrix are reconstructed based on a graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; a directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network. The network reconstruction method ensures the effectiveness of retaining important edges in the regulatory network, which can improve the effectiveness of identifying complex interactions between genes, greatly improve the accuracy of distinguishing direct and indirect regulatory relationships, meet the dynamic identification needs of adaptive biological tissue gene regulation, and thus improve the identification accuracy of transcription factors.

[0021] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present application. The same reference symbols are used throughout the drawings to represent the same components. In the drawings: Figure 1 A flow chart of a method for identifying transcription factors of tissue genes provided in an embodiment of the present application is shown; Figure 2 A schematic diagram of a directed gene regulatory network reconstruction process provided by an embodiment of the present application is shown; Figure 3 A block diagram of a device for identifying transcription factors of tissue genes provided in an embodiment of the present application is shown; Figure 4 A schematic diagram of the structure of a terminal provided in an embodiment of the present application is shown. DETAILED DESCRIPTION

[0023] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0024] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects and are not necessarily used to describe a specific order or sequential order. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present application described herein can be implemented in a sequence other than those illustrated or described herein. In addition, the terms "including" and "having" and any of their variations are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device comprising a series of steps or units is not necessarily limited to those steps or units clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0025] The present application provides a method for identifying transcription factors of tissue genes, such as Figure 1 As shown, the method includes: 101. Obtain an initial gene expression matrix of tissue genes, and construct a fully connected directed graph based on the initial gene expression matrix.

[0026] In the present application embodiment, the current execution subject can be a terminal device or a cloud server as the processing end for identifying the transcription factor of the tissue gene, so as to obtain the initial gene expression matrix of the tissue gene. Wherein, the tissue gene can be the gene content of the tissue cell obtained by different species through different dimensions in the large-scale spatiotemporal omics data set, or the gene content of the tissue cell collected in real time, and the present application embodiment is not specifically limited. The initial gene expression matrix is ​​used to characterize the matrix of the tissue gene in cell dimension and feature dimension, and is a description form of the matrix representation of the tissue gene using the mathematical field, and the present application embodiment is not specifically limited. In addition, the fully connected directed graph is the directed structure graph comprising gene node and potential regulation edge, and specifically, gene can be represented as the node in the directed structure graph, and potential regulation relationship is represented as the edge in the directed structure graph, to build a fully connected directed graph.

[0027] 102. After converting the fully connected directed graph into an undirected role graph, reconstruct the undirected role graph and the feature matrix based on the graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix.

[0028] In an embodiment of the present application, after the current execution end obtains a fully connected directed graph, in order to be able to serve as the input of the neural network model, the fully connected directed graph is converted to obtain an undirected role graph, wherein the edges in the fully connected directed graph are directed and are used to characterize the potential regulatory relationships with directionality between different gene nodes. Correspondingly, the edges in the undirected role graph are non-directional and are used to characterize the potential regulatory relationships without directionality between different gene nodes. This embodiment of the present application does not make specific limitations. Furthermore, the current execution end reconstructs based on the undirected role graph and the feature matrix to obtain a reconstructed target gene expression matrix. At this time, the feature matrix is ​​used to characterize the matrix content derived spatially from the expression data of tissue genes at gene sites or cells, and is reconstructed in combination with the adjacency matrix corresponding to the undirected role graph (obtained by converting the undirected role graph) as the input of the graph neural network model. The loss value of the graph neural network model is calculated after dynamically removing edges. That is, during the model training process, the loss value obtained by the training model is calculated by deleting each edge in the graph one by one. At this time, dynamic edge removal means that an edge is selected for deletion during each round of training, and the loss value of this round is calculated until the loss value meets the preset conditions, and the model training is completed. The graph neural network model is a two-layer graph neural network GNN (Graph Neural Network). The graph neural network model consists of nodes (such as tissue genes) and edges (such as regulatory relationships), which can be represented by an adjacency matrix. This embodiment of the present application does not make specific limitations.

[0029] 103. Generate a directed gene regulatory network based on the target gene expression matrix, and identify the transcription factors of the tissue genes based on the directed gene regulatory network.

[0030] In an embodiment of the present application, after the current execution end obtains the reconstructed target gene expression matrix, it generates a target gene expression matrix, and then uses the optimized target gene expression matrix to generate a directed network graph, namely a directed gene regulatory network GRN (Gene Regulatory Network). Furthermore, based on this directed gene regulatory network, the transcription factors of the tissue genes are identified, thereby improving the accuracy of identifying transcription factors in the biological field. Among them, when generating a directed gene regulatory network based on the target gene expression matrix, the same method as that of constructing a fully connected directed graph using the initial gene expression matrix can be used to construct a directed gene regulatory network containing nodes (genes) and edges (potential regulatory relationships).

[0031] In another embodiment of the present application, for further explanation and limitation, after converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph based on the graph neural network model for which model training has been completed to obtain the reconstructed target gene expression matrix, the method further includes: Get a sample dataset; Build a two-layer graph neural network; During the model training of the two-layer graph neural network based on the sample data set, all network edges are traversed, and when the target network edge is deleted to determine that the loss value is minimum, the model training of the two-layer graph neural network is completed to obtain a graph neural network model that has completed model training.

[0032] In order to optimize the potential regulatory relationship of genes based on the graph neural network model, thereby improving the accuracy of identifying transcription factors, the current execution end first obtains a sample dataset, which includes an undirected role graph sample and a feature matrix sample. At this time, the feature matrix sample can be obtained from the SRT database, which includes data such as STOmics, SOAR, SpatialDB, CROST, 10x Genomics website (available through 10x Genomics), and Census (available through Cellxgene). For example, it contains 96,700,729 cells / points from 7,367 tissue sections, covering 365 tissue cells, including lung, skin, brain, liver, kidney, spinal cord and embryo, and covers normal tissues and various diseases, such as pancreatic ductal adenocarcinoma, amyotrophic lateral sclerosis, non-small cell lung cancer and hepatocellular carcinoma, which are not specifically limited in this embodiment of the present application. At the same time, a two-layer graph neural network is constructed. At this time, the first layer of the two-layer graph neural network is used to aggregate neighbor node features, and the second layer of the two-layer graph neural network is used to generate high-order feature representations. This embodiment of the present application does not make specific limitations.

[0033] It should be noted that in the process of model training of the two-layer graph neural network based on the sample data set, the loss function used is the minimization of the mean square error (MSE) loss function, which is expressed as follows: ; Where N is the number of samples, is the true value, The output obtained by the model is used to calculate the loss value based on this loss function. Meanwhile, during the training process, all network edges are traversed, and when the loss value is minimized by deleting the target network edge, the model training of the two-layer graph neural network is completed, resulting in a graph neural network model that has completed model training.

[0034] In some embodiments, the encoding process of the graph neural network (GNN) includes: aggregating neighbor node features of the first layer GNN and generating high-order feature representations of the second layer GNN. The aggregated neighbor node features of the first layer GNN are calculated as follows: ;in, is the weight of the first layer network, which is used to linearly transform the input feature matrix to extract preliminary feature representation. The nodes obtained for the first layer characteristics. is the activation function (ReLU function can be preferred as the activation function), i is (no i), v is the network node, j is the neighbor node, is the feature vector of the node, is the feature vector of the neighbor node, is the set of neighbor nodes, is the aggregation function, which takes the element-by-element average of the feature vectors of all neighboring nodes as the aggregation operation, as expressed as: The second layer of GNN generates high-order feature representation, and the calculation formula is: ; is the weight of the second layer network, The nodes obtained for the second layer Finally, the output is expressed as In one embodiment of the present application, as the input of the graph neural network model, the feature matrix X, for each node v, the model first extracts the feature matrix from its neighbor set Get the message from the node and then use the aggregation function to The features of the graph neural network model are integrated with the features of the neighboring nodes. At this time, the activation function of the graph neural network model is preferably the ReLU function, and the loss function is preferably the mean square error MSE function (expressed as ), and then, the gradient of the loss function with respect to the model parameters is calculated through the back-propagation algorithm, and the parameters are updated using the Adam optimizer, such as setting the learning rate to 0.0001, so as to minimize the loss value.

[0035] In another embodiment of the present application, for further explanation and limitation, the step traverses all network edges, and when the target network edge is deleted to determine that the loss value is minimized, completes the model training of the two-layer graph neural network, and obtains the graph neural network model that completes the model training, including: Determine the initial loss baseline value; Deleting a target network edge, and calculating a loss value of the two-layer graph neural network corresponding to deleting the target network edge; When the loss value is less than the initial loss reference value, the initial loss reference value is updated based on the loss value, and the network edge is deleted; When the loss value is greater than or equal to the initial loss reference value, the target network edge is updated, and the step of calculating and deleting the loss value of the two-layer graph neural network corresponding to the target network edge is re-executed.

[0036] In order to preserve the key regulatory relationships and optimize the regulatory network structure, thereby improving the network reconstruction accuracy, the current execution end first determines an initial loss benchmark value when training the model, which is used as the initial loss benchmark value. The configuration can be based on the loss value obtained from the first training model, or based on the experience value during the actual measurement process. This embodiment of the application does not make specific restrictions. As the target network edge, delete it to calculate the loss value of the two-layer graph neural network corresponding to deleting this target network edge, that is, when deleting the edge Retrain the model based on the graph structure after the loss to get a new loss value When the loss value Less than the initial loss benchmark value When the edge is deleted The network model after , has a better fit to the data. Deleting this edge will make the network structure more suitable for the real GRN. Therefore, the initial loss baseline value is set to The value of is updated to When the loss value Greater than or equal to the initial loss benchmark value When It contributes significantly to the performance of the model, so this edge is retained. , update the target network edges to evaluate the impact of each edge on the model, and then delete the updated target network edges, thereby re-executing the steps of calculating the loss value of the two-layer graph neural network corresponding to the deletion of the updated target network edges until the model optimization is completed.

[0037] It should be noted that in order to ensure that the neural network model has the ability to remove noise connections, the current execution end calculates the loss change during the new round of model training after deleting the edge. .like <0, indicating that the loss is reduced, then this edge is determined to be redundant and deleted. >0, indicating an increase in loss, and it is retained as a key regulatory edge, ultimately generating a sparse network with both predictive accuracy and biological significance, effectively eliminating noise connections to highlight the core regulatory relationship.

[0038] In another embodiment of the present application, for further explanation and limitation, after the step of constructing a fully connected directed graph based on the initial gene expression matrix, the method further includes: Splitting the nodes in the fully connected directed graph to obtain input nodes and output nodes; The input node and the output node are encoded based on a bidirectional encoder to obtain the undirected role graph and the feature matrix.

[0039] In order to convert the undirected graph to a graph neural network-based reconstruction, the current execution end first splits the nodes in the fully connected directed graph. At this time, the node represents a gene. When splitting, the node corresponding to each gene can be split into a "source node", i.e., an input node, and a "target node", i.e., an output node, while retaining the regulatory direction information. Specifically, the fully connected directed graph is represented as , V is a node, E is an edge. Since the nodes in a directed graph are directed, when converted to an undirected graph, the nodes Converted into two nodes, namely and , Directed edges in exist Mapped as undirected edges , n is the total number of nodes in the graph neural network.

[0040] In some embodiments, as Figure 2 As shown, after the conversion, the network middle, Directed edges The corresponding representation in an undirected graph is In addition, the obtained undirected graph can also be converted back to a directed graph to maintain the integrity and biological significance of the gene regulatory network. A reverse conversion method can be used, which is not specifically limited in the present embodiment. For the undirected graph G2 output by the graph neural network model, the undirected edges Represents an edge in a directed graph , a is the node, bn is the target node.

[0041] In another embodiment of the present application, for further explanation and limitation, before the step of obtaining the initial gene expression matrix of tissue genes, the method further includes: Determine RNA sequence data of tissue genes; Deleting the RNA sequence data of low-expressed genes and abnormal cell genes from the RNA sequence data based on a preset pipeline, and normalizing and standardizing the deleted RNA sequence data to obtain gene data; The initial gene expression matrix is ​​constructed based on the gene data.

[0042] In order to effectively construct a regulatory network based on the graph structure and the characteristics of the tissue genes to improve the effectiveness of transcription factor identification, the current execution end predetermines the RNA sequence data of the tissue genes. Among them, the RNA sequence data of the tissue genes can be single-cell RNA sequencing data or batch cell RNA sequencing data. The preset pipeline is the Scanpy pipeline in the field of bioinformatics, which is used to eliminate low-expressed genes and abnormal cells, and to normalize and standardize the deleted RNA sequence data to obtain gene data. At this time, logarithmic normalization and Z-score standardization can be used for processing to obtain gene data. For example, more than three cell expressions are finally obtained, or cells expressing more than 200 genes are retained, that is, only low-expressed genes and abnormal cells are eliminated to construct an initial gene expression matrix based on this gene data. Among them, when constructing the initial gene expression matrix, the gene data is used to construct a matrix of cell and feature dimensions, that is, row i in the matrix is ​​a cell, column j is a gene, and element v is gene expression. This embodiment of the present application is not specifically limited.

[0043] In another embodiment of the present application, for further explanation and limitation, the step of constructing a fully connected directed graph based on the initial gene expression matrix includes: The network edges in the fully connected directed graph are screened based on the Pearson correlation coefficient or the protein interaction coefficient to obtain a fully connected directed graph to be converted into an undirected graph.

[0044] In order to make the structural graph with potential regulatory relationships more effective, after the current execution end is initially converted into a fully connected directed graph, the network edges in the converted fully connected directed graph can be screened by the Pearson correlation coefficient or the protein interaction coefficient to obtain the optimal fully connected directed graph. Among them, the Pearson Correlation Coefficient (PearsonCorrelation Coefficient) is an indicator used in statistics to measure the degree of linear correlation between two continuous variables. In the embodiment of the present application, it can be pre-configured based on the characteristics of genes in different tissues, or configured based on scenario requirements, without specific limitations. The protein interaction coefficient (Protein-Protein Interaction Coefficient, PPIC) is an indicator that quantifies the binding strength or interaction probability between proteins. In the embodiment of the present application, it can be configured through actual measurement or based on scenario requirements, without specific limitations.

[0045] In another embodiment of the present application, for further illustration and limitation, the step of identifying the transcription factor of the tissue gene based on the directed gene regulatory network includes: determining the out-degree of the tissue gene according to the directed gene regulatory network; By comparing the out-degrees horizontally, the maximum out-degree in the directed gene regulatory network is determined, and the transcription factor of the tissue gene is determined based on the cell type corresponding to the maximum out-degree.

[0046] To ensure effective identification of transcription factors, the current execution end determines the out-degree of the tissue gene based on the obtained directed gene regulatory network. At this time, out-degree is an indicator used to describe the interaction relationship between a gene and other genes or molecules. It can be determined based on the number of other genes or pathways regulated by a gene in the directed gene regulatory network. Specifically, after calculating the out-degree of a tissue gene, the out-degrees of all genes are compared horizontally. By comparing the out-degrees horizontally, the maximum out-degree in the directed gene regulatory network is determined, and the tissue gene corresponding to the maximum out-degree and the cell type corresponding to this tissue gene are determined. At this time, since the gene out-degree (Out-degree) in the obtained gene regulatory network represents the number of interactions between the gene and other genes or molecules, genes with high out-degrees are usually hub nodes in the gene regulatory network and play a key role in specific cell types. Therefore, the difference in genes is the molecular basis of cell differentiation, and then through selective transcription and translation regulation, differentiated gene expression profiles are formed to determine different cell types. In the embodiment of the present application, the cell type corresponding to the gene with the highest out-degree is the cell type with the most active regulation of the gene. Therefore, it plays an important role in the development or functional maintenance of this cell type, that is, as a cell type-specific transcription factor.

[0047] It should be noted that the method for determining the cell type can be through single-cell sequencing technology, that is, identifying the cell type through Marker genes (marker genes). For example, databases such as CellMarker and PanglaoDB provide query bases for marker genes of different cell types. It can also be queried through tools such as ScType and SingleR, that is, gene expression data can be used to automatically identify cell types. This is not specifically limited in the embodiments of the present application.

[0048] In a specific embodiment, Figure 2 As shown in (a) in the figure, the current execution end constructs a fully connected directed graph from the cell gene expression as the initial gene expression matrix, that is, as a correlation network, and converts it through a bidirectional vector encoding method to obtain an undirected role graph and a feature matrix. Figure 2 (b) in the Figure 2 As shown in (d), the gene ... Split it to get the "source node", that is, the input node, and the "target node", that is, the output node, corresponding to ... ,as well as ... , where each node Can be input by the node and output nodes Then, the feature matrix and the undirected role graph are reconstructed until the loss value is minimized, and a directed gene regulatory network GRN is obtained, such as Figure 2 As shown in (c) in the figure, during the training process, the loss value is calculated after deleting any edge, so that the model training is completed when the loss value meets the conditions.

[0049] An embodiment of the present application provides a method for identifying transcription factors of tissue genes. Compared with the prior art, the embodiment of the present application obtains the initial gene expression matrix of the tissue genes and constructs a fully connected directed graph based on the initial gene expression matrix; after converting the fully connected directed graph into an undirected role graph, the undirected role graph and the feature matrix are reconstructed based on a graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; a directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network. The network reconstruction method ensures the effectiveness of retaining important edges in the regulatory network, which can improve the effectiveness of identifying complex interactions between genes, greatly improve the accuracy of distinguishing direct and indirect regulatory relationships, meet the dynamic identification needs of adaptive biological tissue gene regulation, and thus improve the identification accuracy of transcription factors.

[0050] Furthermore, as a response to the above Figure 1 The present invention provides a method for realizing the transcription factor identification of tissue genes. Figure 3 As shown, the device includes: An acquisition module 21 is used to acquire an initial gene expression matrix of tissue genes and construct a fully connected directed graph based on the initial gene expression matrix; A reconstruction module 22 is configured to, after converting the fully connected directed graph into an undirected role graph, reconstruct the undirected role graph and the feature matrix based on a graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, wherein the loss value of the graph neural network model is calculated after dynamically removing edges; The identification module 23 is configured to generate a directed gene regulatory network based on the target gene expression matrix, and identify the transcription factors of the tissue genes based on the directed gene regulatory network.

[0051] Furthermore, the device further comprises: a construction module, a training module, The acquisition module is further configured to acquire a sample data set, wherein the sample data set includes an undirected role graph sample and a feature matrix sample; The building module is used to construct a two-layer graph neural network, wherein the first layer of the two-layer graph neural network is used to aggregate neighbor node features, and the second layer of the two-layer graph neural network is used to generate high-order feature representations; The training module is used to traverse all network edges during the model training of the two-layer graph neural network based on the sample data set, and complete the model training of the two-layer graph neural network when the target network edge is deleted to determine that the loss value is minimized, so as to obtain a graph neural network model that has completed model training.

[0052] Furthermore, the training module is specifically used to determine an initial loss benchmark value; delete the target network edge, and calculate the loss value of the two-layer graph neural network corresponding to the deletion of the target network edge; when the loss value is less than the initial loss benchmark value, the initial loss benchmark value is updated based on the loss value, and the network edge is deleted; when the loss value is greater than or equal to the initial loss benchmark value, the target network edge is updated, and the step of calculating the loss value of the two-layer graph neural network corresponding to the deletion of the target network edge is re-executed.

[0053] Furthermore, the device further comprises: A splitting module, configured to split the nodes in the fully connected directed graph to obtain input nodes and output nodes; An encoding module is used to encode the input node and the output node based on a bidirectional encoder to obtain the undirected role graph and the feature matrix.

[0054] Furthermore, the device further includes: a determination module, a processing module, The determination module is used to determine the RNA sequence data of the tissue gene; The processing module is used to delete the RNA sequence data of low-expressed genes and abnormal cell genes in the RNA sequence data based on a preset pipeline, and normalize and standardize the deleted RNA sequence data to obtain gene data; The construction module is further used to construct the initial gene expression matrix based on the gene data.

[0055] Furthermore, the construction module is also used to screen the network edges in the fully connected directed graph based on the Pearson correlation coefficient or the protein interaction coefficient to obtain a fully connected directed graph to be converted into an undirected graph.

[0056] Furthermore, the identification module is specifically used to determine the out-degree of the tissue gene based on the directed gene regulatory network; determine the maximum out-degree in the directed gene regulatory network by horizontally comparing the out-degrees, and determine the transcription factor of the tissue gene based on the cell type corresponding to the maximum out-degree.

[0057] An embodiment of the present application provides a transcription factor identification device for tissue genes. Compared with the prior art, the embodiment of the present application obtains the initial gene expression matrix of the tissue genes and constructs a fully connected directed graph based on the initial gene expression matrix; after converting the fully connected directed graph into an undirected role graph, the undirected role graph and the feature matrix are reconstructed based on a graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; a directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network. The network reconstruction method ensures the effectiveness of retaining important edges in the regulatory network, which can improve the effectiveness of identifying complex interactions between genes, greatly improve the accuracy of distinguishing direct and indirect regulatory relationships, meet the dynamic identification needs of adaptive biological tissue gene regulation, and thus improve the identification accuracy of transcription factors.

[0058] According to one embodiment of the present application, a storage medium is provided, wherein the storage medium stores at least one executable instruction. The computer-executable instruction can execute the method for identifying transcription factors of tissue genes in any of the above method embodiments.

[0059] Figure 4 A schematic diagram of the structure of a terminal provided according to an embodiment of the present application is shown. The specific embodiment of the present application does not limit the specific implementation of the terminal.

[0060] like Figure 4 As shown, the terminal may include: a processor 302 , a communications interface 304 , a memory 306 , and a communication bus 308 .

[0061] The processor 302 , the communication interface 304 , and the memory 306 communicate with each other via a communication bus 308 .

[0062] The communication interface 304 is used to communicate with other devices such as clients or other servers.

[0063] The processor 302 is configured to execute the program 310 , and specifically to execute the relevant steps in the above-mentioned embodiment of the method for identifying transcription factors of tissue genes.

[0064] Specifically, the program 310 may include program codes, which include computer operation instructions.

[0065] Processor 302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of the present application. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs, or processors of different types, such as one or more CPUs and one or more ASICs.

[0066] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, or may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0067] The program 310 may be specifically configured to cause the processor 302 to perform the following operations: Obtaining an initial gene expression matrix of tissue genes, and constructing a fully connected directed graph based on the initial gene expression matrix; After converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph and the feature matrix based on the graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; A directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network.

[0068] Obviously, those skilled in the art should understand that the modules or steps of the present application described above can be implemented using a general-purpose computing device, they can be concentrated on a single computing device, or distributed on a network composed of multiple computing devices. Alternatively, they can be implemented using program code executable by the computing device, so that they can be stored in a storage device and executed by the computing device. In some cases, the steps shown or described can be performed in a different order than herein, or they can be made into separate integrated circuit modules, or multiple modules or steps can be made into a single integrated circuit module for implementation. Thus, the present application is not limited to any specific combination of hardware and software.

[0069] The above description is merely a preferred embodiment of the present application and is not intended to limit the present application. Persons skilled in the art will readily appreciate that various modifications and variations are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present application shall be included within the scope of protection of the present application.

Claims

1. A method for identifying transcription factors of tissue genes, characterized in that: include: Obtaining an initial gene expression matrix of tissue genes, and constructing a fully connected directed graph based on the initial gene expression matrix; After converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph and the feature matrix based on the graph neural network model that has completed model training to obtain a reconstructed target gene expression matrix, and the loss value of the graph neural network model is calculated after dynamically removing edges; A directed gene regulatory network is generated based on the target gene expression matrix, and the transcription factors of the tissue genes are identified based on the directed gene regulatory network.

2. The method according to claim 1, characterized in that After converting the fully connected directed graph into an undirected role graph, reconstructing the undirected role graph based on a graph neural network model for which model training has been completed to obtain a reconstructed target gene expression matrix, the method further includes: Acquire a sample data set, wherein the sample data set includes an undirected role graph sample and a feature matrix sample; Constructing a two-layer graph neural network, wherein the first layer of the two-layer graph neural network is used to aggregate neighbor node features, and the second layer of the two-layer graph neural network is used to generate high-order feature representations; During the model training of the two-layer graph neural network based on the sample data set, all network edges are traversed, and when the target network edge is deleted to determine that the loss value is minimum, the model training of the two-layer graph neural network is completed to obtain a graph neural network model that has completed model training.

3. The method according to claim 2, characterized in that The traversal of all network edges and the completion of model training of the two-layer graph neural network when the target network edge is deleted to determine that the loss value is the minimum, and the graph neural network model that completes the model training includes: Determine the initial loss baseline value; Deleting a target network edge, and calculating a loss value of the two-layer graph neural network corresponding to deleting the target network edge; When the loss value is less than the initial loss reference value, the initial loss reference value is updated based on the loss value, and the network edge is deleted; When the loss value is greater than or equal to the initial loss reference value, the target network edge is updated, and the step of calculating and deleting the loss value of the two-layer graph neural network corresponding to the target network edge is re-executed.

4. The method according to claim 1, wherein After constructing the fully connected directed graph based on the initial gene expression matrix, the method further includes: Splitting the nodes in the fully connected directed graph to obtain input nodes and output nodes; The input node and the output node are encoded based on a bidirectional encoder to obtain the undirected role graph and the feature matrix.

5. The method according to claim 1, wherein Before obtaining the initial gene expression matrix of tissue genes, the method further comprises: Determine RNA sequence data of tissue genes; Deleting the RNA sequence data of low-expressed genes and abnormal cell genes from the RNA sequence data based on a preset pipeline, and normalizing and standardizing the deleted RNA sequence data to obtain gene data; The initial gene expression matrix is ​​constructed based on the gene data.

6. The method according to claim 1, characterized in that The constructing of a fully connected directed graph based on the initial gene expression matrix comprises: The network edges in the fully connected directed graph are screened based on the Pearson correlation coefficient or the protein interaction coefficient to obtain a fully connected directed graph to be converted into an undirected graph.

7. The method according to any one of claims 1 to 6, characterized in that The method for identifying the transcription factor of the tissue gene based on the directed gene regulatory network includes: determining the out-degree of the tissue gene according to the directed gene regulatory network; By comparing the out-degrees horizontally, the maximum out-degree in the directed gene regulatory network is determined, and the transcription factor of the tissue gene is determined based on the cell type corresponding to the maximum out-degree.

8. A device for identifying transcription factors of tissue genes, characterized in that: include: an acquisition module, configured to acquire an initial gene expression matrix of tissue genes and construct a fully connected directed graph based on the initial gene expression matrix; A reconstruction module is used to reconstruct the undirected role graph and the feature matrix based on the graph neural network model that has completed model training after converting the fully connected directed graph into an undirected role graph to obtain a reconstructed target gene expression matrix, wherein the loss value of the graph neural network model is calculated after dynamically removing edges; An identification module is used to generate a directed gene regulatory network based on the target gene expression matrix, and identify the transcription factors of the tissue genes based on the directed gene regulatory network.

9. A computer-readable storage medium having a computer program / instruction stored thereon, characterized in that: When the computer program / instructions are executed by a processor, the steps of the method according to claim 1 are implemented.

10. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to claim 1.

Images