A method, system and medium for constructing a liver cancer gene database

Abstract

The present invention relates to the field of medical information technology, in particular to a construction method, system and medium of a liver cancer gene database. The method constructs a spatial neighborhood relationship based on cell spatial coordinates, and defines a group of interconnected cells as a spatial neighborhood hyperedge; annotates functional modules according to gene expression values, and associates cell nodes within the same functional module as functional module hyperedges; constructs a spatio-temporal dynamic hypergraph structure based on the spatial neighborhood hyperedges and functional module hyperedges; performs local aggregation on the node features of the spatio-temporal dynamic hypergraph structure to obtain a gene expression correlation weight matrix between a target node and domain nodes, thereby constructing a liver cancer gene database. The present invention comprehensively depicts the spatial interaction network, the dynamic evolution of functional modules and the gene expression correlation of liver cancer cells, providing high-quality data support and an analysis platform for the study of liver cancer mechanisms.

Description

Technical Field

[0001] The present invention relates to the field of medical information technology, and particularly to a method, system, and medium for constructing a liver cancer gene database. Background Art

[0002] With the development of spatial transcriptomics technology, researchers can simultaneously obtain the gene expression information and its spatial distribution characteristics of single cells in liver cancer tissues, which provides an important data basis for revealing the spatial heterogeneity of the tumor microenvironment and cell-cell interactions. However, most existing liver cancer gene databases are limited to the static storage of gene expression profiles or clinical information, lacking the integrated analysis of spatial neighborhood relationships, dynamic evolution of functional modules, and gene expression correlations between cells. In addition, traditional databases have the following limitations in data processing: First, they fail to fully utilize spatial coordinate information to construct cell neighborhood relationships, resulting in the lack of a spatial interaction network; second, they lack dynamic tracking of functional modules and are difficult to capture the temporal changes of the tumor microenvironment; third, they do not introduce a hypergraph structure to uniformly model multi-modal data, limiting the expression and analysis capabilities of complex relationships. In view of this, the present invention proposes a method, system, and medium for constructing a liver cancer gene database that can integrate spatial transcriptome data, construct a spatio-temporal dynamic hypergraph structure, and support multi-dimensional queries. Summary of the Invention

[0003] The present invention overcomes the deficiencies of the prior art and provides a method, system, and medium for constructing a liver cancer gene database.

[0004] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0005] In the first aspect of the present invention, a method for constructing a liver cancer gene database is disclosed, including the following steps:

[0006] Obtain spatial transcriptome data of liver cancer tissues, including gene expression values, cell spatial coordinates, and time series sampling data;

[0007] Based on the cell spatial coordinates, construct spatial neighborhood relationships, define interconnected cell groups as spatial neighborhood hyperedges, and obtain the spatial attributes of the spatial neighborhood hyperedges;

[0008] Label functional modules according to gene expression values, associate cell nodes within the same functional module as functional module hyperedges, and obtain the representative features of the functional module hyperedges;

[0009] Based on the spatial neighborhood hyperedges and functional module hyperedges, construct a spatio-temporal dynamic hypergraph structure, and obtain the data units of the spatio-temporal dynamic hypergraph structure;

[0010] Perform local aggregation on the node features of the spatio-temporal dynamic hypergraph structure to obtain the gene expression correlation weight matrix between the target node and the neighborhood nodes;

[0011] Construct a liver cancer gene database based on the spatial attributes, representative features, data units, and gene expression correlation weight matrix.

[0012] Preferably, construct a spatial neighborhood relationship based on the cell spatial coordinates, define the interconnected cell groups as spatial neighborhood hyperedges, and obtain the spatial attributes of the spatial neighborhood hyperedges, specifically:

[0013] Based on the cell spatial coordinates of the cells, connect each cell into a triangle by the Delaunay triangulation method to generate a triangular mesh graph;

[0014] Traverse all the triangles in the triangular mesh graph, mark the two cells sharing the same edge as having a spatial neighborhood relationship, and construct an adjacency matrix between the cells;

[0015] Based on the adjacency matrix, obtain all the interconnected cell groups. If a group of cells is interconnected through neighborhood relationships, then define this group of cells as a spatial neighborhood hyperedge, and record the unique identifiers of all the cells within the spatial neighborhood hyperedge;

[0016] Label spatial attributes for each spatial neighborhood hyperedge, including the hyperedge center coordinates, coverage area, number of cells, and hyperedge type.

[0017] It also includes:

[0018] If the number of cells within a certain spatial neighborhood hyperedge is less than a preset value, then determine it as an isolated hyperedge and eliminate it; if the spatial distribution density of the cells within a certain spatial neighborhood hyperedge is lower than a preset density threshold, then determine it as a sparse hyperedge;

[0019] For a sparse hyperedge, recalculate its Euclidean distance from the adjacent hyperedges. If the Euclidean distance is less than a preset merging threshold, then merge this sparse hyperedge with the adjacent hyperedge into a new spatial neighborhood hyperedge, and update the list of unique identifiers of the cells within the hyperedge.

[0020] Preferably, label the functional modules according to the gene expression values, associate the cell nodes within the same functional module as a functional module hyperedge, and obtain the representative features of the functional module hyperedge, specifically:

[0021] Calculate the gene expression similarity between cells based on the gene expression values to generate a symmetric similarity matrix;

[0022] Based on the similarity matrix, adopt the hierarchical clustering method to recursively merge the most similar cells or cell clusters, gradually generate a clustering hierarchy until all cells are grouped into a single cluster, and obtain a cell clustering dendrogram;

[0023] Calculate the silhouette coefficient of each clustering module in the cell clustering dendrogram; if the silhouette coefficient of a certain clustering module is greater than the preset coefficient threshold, then divide this module into an independent co-expression gene module and record the unique identifiers of all cells within the module;

[0024] Perform functional enrichment analysis on each co-expression gene module; if the enrichment significance of a certain co-expression gene module is lower than the preset significance threshold, then label it as a functional module;

[0025] Associate the cell nodes within the same functional module as functional module hyperedges, record the unique identifiers of the cells within the functional module hyperedges and the functional module type, and at the same time use the central expression profile of the functional module hyperedge as its representative feature.

[0026] Preferably, construct a spatio-temporal dynamic hypergraph structure based on the spatial neighborhood hyperedges and functional module hyperedges, and obtain the data units of the spatio-temporal dynamic hypergraph structure, specifically:

[0027] Divide the time series sampling data into multiple time slices based on the time series sampling points, and each slice contains the gene expression values, cell spatial coordinates, and hyperedge connection status at the current time point;

[0028] In the first time slice, construct a hypergraph structure based on the spatial neighborhood hyperedges and functional module hyperedges, and record the connection relationships between the hypergraph nodes and the spatial neighborhood hyperedges and functional module hyperedges;

[0029] For subsequent time slices, traverse all cell nodes. If their spatial coordinates or gene expression values change, then recalculate their affiliated hyperedges and update the hyperedge connection status;

[0030] If a hyperedge remains unchanged in consecutive time slices, then label it as a "persistent hyperedge"; if a hyperedge disappears in a certain time slice, then label it as a "disappeared hyperedge"; if a hyperedge appears for the first time in a certain time slice, then label it as a "newly added hyperedge";

[0031] If a persistent hyperedge shows similar gene expression patterns or spatial distribution characteristics in several time slices, then label it as a "spatio-temporal stable hyperedge" and store its evolution path;

[0032] If a disappeared hyperedge reappears in subsequent time slices, then restore it to a "persistent hyperedge"; if a newly added hyperedge does not appear again in subsequent time slices, then label it as a "temporary hyperedge" and eliminate it;

[0033] Store the hypergraph structure of each time slice as an independent data unit, and the data unit contains the hyperedge connection status, cell node attributes, and hyperedge evolution events.

[0034] Preferably, local aggregation is performed on the node features of the spatio-temporal dynamic hypergraph structure to obtain the gene expression correlation weight matrix between the target node and the domain nodes, specifically as follows:

[0035] Extract the gene expression features of cell nodes in each time slice from the spatio-temporal dynamic hypergraph structure, including gene expression values, spatial coordinates, and the information of the superedges they belong to;

[0036] For each cell node, based on the spatial neighborhood superedges and functional module superedges it belongs to, define its local neighborhood range. If a node belongs to multiple superedges at the same time, extend its neighborhood range to the cell nodes within all relevant superedges;

[0037] Within the local neighborhood range, calculate the gene expression similarity between the target node and the neighborhood nodes; if the similarity value is greater than the preset similarity threshold, mark the target node and the neighborhood nodes as high-similarity node pairs;

[0038] If it is a high-similarity node pair, assign the first weight value; otherwise, assign the second weight value or zero value; obtain the gene expression correlation weight matrix; where the first weight value is greater than the second weight value.

[0039] Preferably, construct a liver cancer gene database according to the spatial attributes, representative features, data units, and the gene expression correlation weight matrix, specifically as follows:

[0040] Integrate the spatial attributes of the spatial neighborhood superedges, the representative features of the functional module superedges, the data units of the spatio-temporal dynamic hypergraph structure, and the gene expression correlation weight matrix to form liver cancer gene feature data;

[0041] Use a graph database as the storage architecture, with hypergraph nodes as entities and spatial neighborhood relationships, time evolution paths, and gene expression correlations as edges to construct an interaction relationship map of liver cancer genes and space;

[0042] Store the liver cancer gene feature data in the interaction relationship map in layers according to time slices to form a liver cancer gene database; where each time slice corresponds to an independent data layer, supporting query and visualization display by time point;

[0043] When new liver cancer gene feature data is added, only update the affected nodes and their associated information. If a node appears in the new time slice, generate a new node information unit and associate it with the interaction relationship map; if a node disappears, mark it as a "historical node" and retain its evolution path.

[0044] In a second aspect of the present invention, a construction system for a liver cancer gene database is disclosed. The construction system for the liver cancer gene database includes a memory and a processor. A program for the construction method of the liver cancer gene database is stored in the memory. When the program for the construction method of the liver cancer gene database is executed by the processor, the steps of any one of the construction methods of the liver cancer gene database are implemented.

[0045] In a third aspect of the present invention, a computer-readable storage medium is disclosed. The computer-readable storage medium includes a program for the construction method of the liver cancer gene database. When the program for the construction method of the liver cancer gene database is executed by the processor, the steps of any one of the construction methods of the liver cancer gene database are implemented.

[0046] The present invention solves the technical defects in the background art and has the following beneficial effects: By obtaining the spatial transcriptome data of liver cancer tissues, constructing spatial neighborhood hyperedges and functional module hyperedges, and integrating multi-dimensional features based on the spatio-temporal dynamic hypergraph structure, the present invention comprehensively depicts the spatial interaction network of liver cancer cells, the dynamic evolution of functional modules, and the gene expression correlation, thereby constructing a liver cancer gene database and providing high-quality data support and an analysis platform for the study of liver cancer mechanisms. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the following-described drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0048] Figure 1 It is a first method flow chart of a construction method for a liver cancer gene database;

[0049] Figure 2 It is a second method flow chart of a construction method for a liver cancer gene database;

[0050] Figure 3 It is a system block diagram of a construction system for a liver cancer gene database. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0051] In order to be able to more clearly understand the above objects, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that, without conflict, the embodiments of the present application and the features in the embodiments can be combined with each other.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. However, the present invention may also be implemented in other ways different from those described herein. Therefore, the scope of protection of the present invention is not limited by the specific embodiments disclosed below.

[0053] As Figure 1 shown, a method for constructing a liver cancer gene database according to a first aspect of the present invention includes the following steps:

[0054] S102. Obtain spatial transcriptome data of liver cancer tissues, including gene expression values, cell spatial coordinates, and time-series sampling data;

[0055] S104. Construct a spatial neighborhood relationship based on cell spatial coordinates, define a group of interconnected cells as a spatial neighborhood hyperedge, and obtain the spatial attributes of the spatial neighborhood hyperedge;

[0056] S106. Label functional modules according to gene expression values, associate cell nodes within the same functional module as a functional module hyperedge, and obtain the representative features of the functional module hyperedge;

[0057] S108. Construct a spatio-temporal dynamic hypergraph structure based on the spatial neighborhood hyperedge and the functional module hyperedge, and obtain the data units of the spatio-temporal dynamic hypergraph structure;

[0058] S110. Perform local aggregation on the node features of the spatio-temporal dynamic hypergraph structure to obtain the gene expression correlation weight matrix between the target node and the neighborhood nodes;

[0059] S112. Construct a liver cancer gene database according to the spatial attributes, representative features, data units, and gene expression correlation weight matrix.

[0060] It should be noted that the present invention constructs a spatial neighborhood hyperedge and a functional module hyperedge by obtaining the spatial transcriptome data of liver cancer tissues, and integrates multi-dimensional features based on the spatio-temporal dynamic hypergraph structure to comprehensively depict the spatial interaction network of liver cancer cells, the dynamic evolution of functional modules, and the gene expression correlation, thereby constructing a liver cancer gene database to provide high-quality data support and analysis platform for the study of liver cancer mechanisms.

[0061] Preferably, constructing a spatial neighborhood relationship based on cell spatial coordinates, defining a group of interconnected cells as a spatial neighborhood hyperedge, and obtaining the spatial attributes of the spatial neighborhood hyperedge is specifically as follows:

[0062] Based on the cell spatial coordinates of the cells, connect each cell into a triangle by the Delaunay triangulation method to generate a triangular mesh graph; where the vertices of each triangle are cell coordinate points, used to define the spatial neighborhood relationship between cells;

[0063] It should be noted that the Delaunay triangulation method generates an optimized triangular mesh map by using the cell space coordinate points as vertices to construct a series of non-overlapping triangles, ensuring that no other vertices are contained within the circumcircle of each triangle. By maximizing the minimum angle, the generation of long and narrow triangles is avoided, ensuring the uniformity and stability of the grid, and at the same time accurately reflecting the spatial neighborhood relationship between cells, providing a reliable basis for the subsequent definition and analysis of spatial hyperedges.

[0064] Traverse all the triangles in the triangular mesh map, mark two cells sharing the same edge as having a spatial neighborhood relationship, and construct an adjacency matrix between cells. The element value of 1 in the adjacency matrix indicates that the two cells are in a neighborhood relationship, otherwise it is 0.

[0065] Based on the adjacency matrix, obtain all mutually connected cell groups. If a group of cells is mutually connected through neighborhood relationships, then define this group of cells as a spatial neighborhood hyperedge, and record the unique identifiers of all cells within the spatial neighborhood hyperedge. Here, the identifier refers to a unique number or code assigned to each cell to uniquely identify the cell in the database. The identifier is based on the original data of the cell (such as spatial coordinates, gene expression characteristics) or a serial number automatically generated by the system to ensure that the identity of each cell within the hyperedge is unique and traceable.

[0066] Exemplarily, all mutually connected cell groups can be obtained through the Depth-First Search (DFS) method: starting from a certain cell node, recursively traverse all its neighborhood nodes and mark the visited nodes, thereby obtaining all mutually connected cell groups. By exploring the neighborhood relationship layer by layer, it is ensured that all connected cells within the same hyperedge can be completely identified, while avoiding repeated visits.

[0067] Annotate spatial attributes for each spatial neighborhood hyperedge, including the hyperedge center coordinates, the area of the covered region, the number of cells, and the hyperedge type (such as tumor cell hyperedge, immune cell hyperedge).

[0068] Among them, annotating spatial attributes for each spatial neighborhood hyperedge: First, calculate the mean of all cell spatial coordinates within the hyperedge as the hyperedge center coordinates to ensure that it represents the spatial position of the hyperedge. Secondly, construct a convex hull polygon based on the cell coordinate points and calculate the polygon area as the area of the hyperedge covered region to quantify its spatial range. Then, count the number of cells contained within the hyperedge as a characteristic index of the hyperedge scale. Finally, perform hyperedge type annotation according to the functional types of cells within the hyperedge (such as tumor cells, immune cells). If the proportion of a certain type of cell within the hyperedge exceeds a preset threshold, then label it as the hyperedge of this type, otherwise label it as a mixed type hyperedge.

[0069] It also includes:

[0070] If the number of cells within a spatial neighborhood hyperedge is less than a preset value, it is determined as an isolated hyperedge and removed; if the spatial distribution density of cells within a spatial neighborhood hyperedge is lower than a preset density threshold, it is determined as a sparse hyperedge;

[0071] For a sparse hyperedge, recalculate its Euclidean distance from adjacent hyperedges. If the Euclidean distance is less than a preset merging threshold, merge the sparse hyperedge with the adjacent hyperedge into a new spatial neighborhood hyperedge and update the list of unique identifiers of cells within the hyperedge.

[0072] In summary, this method can characterize the spatial interaction network of cells in liver cancer tissue, effectively identify key neighborhood regions in the tumor microenvironment by constructing spatial neighborhood relationships and defining spatial neighborhood hyperedges based on cell spatial coordinates, and annotating their spatial attributes. And by removing isolated hyperedges and optimizing sparse hyperedges, the integrity and reliability of spatial neighborhood hyperedges are ensured, providing a reliable data basis for revealing the spatial distribution characteristics and functional associations of liver cancer cells.

[0073] Preferably, functional modules are annotated according to gene expression values, cell nodes within the same functional module are associated as functional module hyperedges, and representative features of the functional module hyperedges are obtained, such as Figure 2 shown as follows:

[0074] S202. Calculate the gene expression similarity between cells based on the gene expression values to generate a symmetric similarity matrix, where each element represents the degree of similarity in gene expression patterns between two cells;

[0075] Exemplarily, calculate the gene expression similarity between cells through the cosine similarity algorithm to generate a symmetric similarity matrix: First, vectorize the gene expression values of each cell to ensure consistent vector dimensions; second, calculate the cosine value of the angle between each pair of cell vectors. The formula is the dot product of the two vectors divided by the product of their magnitudes, obtaining a similarity score ranging from [-1, 1]; then, fill the similarity scores of each pair of cells into the corresponding positions of the matrix to generate a symmetric similarity matrix, where the rows and columns of the matrix correspond to cell numbers respectively, the diagonal elements are 1 (indicating complete similarity of a cell to itself), and the remaining elements are the similarity scores between two cells.

[0076] S204. Based on the similarity matrix, use hierarchical clustering to recursively merge the most similar cells or cell clusters to gradually generate a clustering hierarchy until all cells are grouped into a single cluster, obtaining a cell clustering dendrogram;

[0077] It should be noted that based on the similarity matrix, a hierarchical clustering method is used to generate a cell clustering dendrogram. First, each cell is initialized as an independent cluster, and the similarity between all cluster pairs is calculated. Second, the two clusters with the highest similarity are found, merged into a new cluster, and the similarity matrix is updated. The similarity between the new cluster and other clusters is recalculated (such as using the average linkage method). Then, the above merging and updating processes are repeated, gradually recursively merging all cells into larger clusters until all cells are grouped into a single cluster. Finally, the cluster pairs and their similarities for each merge are recorded to generate a cell clustering dendrogram, where the leaf nodes of the dendrogram are individual cells, the internal nodes are the merged clusters, and the branch lengths represent the similarity between clusters.

[0078] Among them, the hierarchical clustering method is an unsupervised clustering algorithm that gradually constructs a hierarchical clustering structure by recursively merging or splitting data or clusters.

[0079] S206. Calculate the silhouette coefficient of each clustering module in the cell clustering dendrogram; if the silhouette coefficient of a certain clustering module is greater than the preset coefficient threshold, then divide this module into an independent co-expression gene module and record the unique identifiers of all cells within the module.

[0080] Among them, the silhouette coefficient is an index used to evaluate the clustering quality, measuring the compactness and separation of clustering modules in the clustering. The silhouette coefficient is a commonly used evaluation index for clustering effects, and the details of how to calculate the silhouette coefficient will not be elaborated here.

[0081] S208. Perform functional enrichment analysis on each co-expression gene module; if the enrichment significance of a certain co-expression gene module is lower than the preset significance threshold, then label it as a functional module.

[0082] Among them, functional enrichment analysis is a statistical method used to identify significantly enriched biological functions or pathways in a gene set. By comparing the target gene set with a known function database (such as GO or KEGG), its enrichment degree is calculated. Its core steps include: first, based on the hypergeometric distribution or Fisher's exact test, calculate the enrichment significance (p-value) of the target gene set in a specific functional category; second, reduce the false positive rate through multiple test corrections (such as FDR); finally, screen the significantly enriched functional categories (such as p-value less than the preset threshold) to provide a basis for revealing the biological significance of the gene set, which is widely used in gene function annotation and module analysis.

[0083] S210. Associate the cell nodes within the same functional module as a functional module hyperedge, record the unique identifiers of the cells within the functional module hyperedge and the functional module type, and at the same time use the central expression profile of the functional module hyperedge as its representative feature.

[0084] The central expression profile of the functional module hyperedge refers to the mean vector of the gene expression values ​​of all cells within the hyperedge, which is used to represent the overall gene expression pattern of the module.

[0085] In summary, this method divides co-expressed gene modules and associates cells in the same functional module into functional module hyperedges, thereby accurately identifying functionally related cell populations in liver cancer tissues and providing visual data for revealing gene co-expression patterns and their biological functions.

[0086] Preferably, a spatiotemporal dynamic hypergraph structure is constructed based on the spatial neighborhood hyperedge and the functional module hyperedge, and a data unit of the spatiotemporal dynamic hypergraph structure is obtained, specifically:

[0087] The time series sampling data is divided into multiple time slices based on the time series sampling points, each slice contains the gene expression value, cell space coordinates and hyperedge connection status at the current time point;

[0088] In the first time slice, a hypergraph structure is constructed based on the spatial neighborhood hyperedges and the functional module hyperedges, and the hyperedge connection relationship between the hypergraph nodes (cells) and the spatial neighborhood hyperedges and the functional modules is recorded;

[0089] It should be noted that, first, each cell is regarded as a hypergraph node and a unique identifier is assigned to it; secondly, all spatial neighborhood hyperedges are traversed, and a connection relationship is established between all cell nodes in the hyperedge and the hyperedge, and the association information between the cell node and the spatial neighborhood hyperedge is recorded; then, all functional module hyperedges are traversed, and a connection relationship is established between all cell nodes in the hyperedge and the hyperedge, and the association information between the cell node and the functional module hyperedge is recorded; finally, the connection relationship between the hypergraph node and the hyperedge is stored as a hypergraph structure.

[0090] For subsequent time slices, all cell nodes are traversed. If their spatial coordinates or gene expression values ​​change, their hyperedges are recalculated and the hyperedge connection status of the hypergraph structure is updated;

[0091] Specifically, if the change of the spatial coordinates of a cell causes it to deviate from the original spatial neighborhood hyperedge, its spatial neighborhood relationship is reconstructed based on the Delaunay triangulation method, and the spatial neighborhood hyperedge is updated; if the change of the cell gene expression value causes it to deviate from the original functional module hyperedge, its functional module is redivided based on the hierarchical clustering method, and the functional module hyperedge is updated.

[0092] If a hyperedge remains unchanged in consecutive time slices, it is marked as a “persistent hyperedge”; if a hyperedge disappears in a certain time slice, it is marked as a “disappearing hyperedge”; if a hyperedge appears for the first time in a certain time slice, it is marked as a “newly added hyperedge”;

[0093] If a persistent hyperedge exhibits similar gene expression patterns or spatial distribution characteristics in several time slices, it is labeled as a "spatiotemporal stable hyperedge", and its evolutionary path is stored;

[0094] If a disappearing hyperedge reappears in subsequent time slices, it is restored to a "persistent hyperedge"; if a newly added hyperedge does not reappear in subsequent time slices, it is marked as a "temporary hyperedge" and removed;

[0095] The hypergraph structure of each time slice is stored as an independent data unit, which contains hyperedge connection status, cell node attributes, and hyperedge evolution events.

[0096] In summary, by constructing a spatiotemporal dynamic hypergraph structure based on time series sampling points, this method can capture the temporal evolution characteristics of cell spatial neighborhood relationships and functional modules in liver cancer tissues; by dynamically updating the hyperedge connection status and labeling "persistent hyperedges", "disappearing hyperedges", and "newly added hyperedges", it reveals the dynamic evolution of the spatial interaction network and functional modules of liver cancer cells.

[0097] Preferably, local aggregation is performed on the node features of the spatiotemporal dynamic hypergraph structure to obtain the gene expression correlation weight matrix between the target node and its neighboring nodes, specifically as follows:

[0098] Extract the gene expression features of cell nodes in each time slice from the spatiotemporal dynamic hypergraph structure, including gene expression values, spatial coordinates, and hyperedge information to which they belong;

[0099] For each cell node, based on the spatial neighborhood hyperedges and functional module hyperedges to which it belongs, define its local neighborhood range. If a node belongs to multiple hyperedges at the same time, its neighborhood range is extended to the cell nodes within all relevant hyperedges;

[0100] It should be noted that all cell nodes within each spatial neighborhood hyperedge and functional module hyperedge are extracted, and these cell nodes are combined into the initial neighborhood range of this node. If a node belongs to multiple hyperedges at the same time, all relevant hyperedges are traversed, and the cell nodes they contain are all included in the neighborhood range to ensure the integrity and extensibility of the neighborhood range; finally, a list of the local neighborhood ranges of this node is generated for subsequent gene expression similarity calculation.

[0101] Within the local neighborhood range, calculate the gene expression similarity between the target node and its neighboring nodes; if the similarity value is greater than the preset similarity threshold, mark the target node and its neighboring nodes as a high-similarity node pair;

[0102] Specifically, the gene expression similarity between the target node and its neighboring nodes is calculated by the cosine similarity algorithm.

[0103] If it is a high similarity node pair, assign the first weight value; otherwise, assign the second weight value or zero; obtain the gene expression correlation weight matrix; where the first weight value is greater than the second weight value.

[0104] Specifically, if the gene expression similarity value of the node pair is greater than the preset similarity threshold (such as 0.8), then mark the node pair as a high similarity node pair and assign the first weight value (such as 1.0); if the gene expression similarity value of the node pair is less than or equal to the preset similarity threshold, then mark the node pair as a low similarity node pair and assign the second weight value (such as 0.5) or zero. By traversing all node pairs, a gene expression correlation weight matrix is generated, where the rows and columns of the matrix correspond to the cell node numbers respectively, and the matrix elements are the weight values of the node pairs. This weight matrix is used to quantify the gene expression correlation between cells.

[0105] In summary, this method captures the gene expression similarity features of liver cancer cells in the spatial neighborhood and functional modules by locally aggregating the node features of the spatio-temporal dynamic hypergraph structure, so as to obtain the gene expression correlation weight matrix between the target node and its neighboring nodes, and to reveal the functional interaction relationship and gene co-expression pattern between cells in the tumor microenvironment.

[0106] Preferably, a liver cancer gene database is constructed according to the spatial attributes, representative features, data units, and gene expression correlation weight matrix, specifically:

[0107] Integrate the spatial attributes of the spatial neighborhood hyperedges (such as central coordinates, covered area, number of cells), the representative features of the functional module hyperedges (such as central expression profile, functional module type), the data units of the spatio-temporal dynamic hypergraph structure (such as hyperedge connection status, evolution path), and the gene expression correlation weight matrix to form liver cancer gene feature data;

[0108] Use a graph database as the storage architecture, with hypergraph nodes (cells) as entities and spatial neighborhood relationships, time evolution paths, and gene expression correlations as edges to construct an interaction relationship map of liver cancer genes and space;

[0109] It should be noted that using a graph database (such as Neo4j) as the storage architecture, first take the hypergraph nodes as entity nodes, assign a unique identifier to each node, and store its gene expression value, spatial coordinates, functional module type and other attributes; use the spatial neighborhood relationship, time evolution path, and gene expression correlation as edges to define the connection relationships between entity nodes respectively, where the spatial neighborhood relationship edge stores the hyperedge type and spatial attributes, the time evolution path edge stores the time slice information and evolution events, and the gene expression correlation edge stores the weight value; finally, integrate the entity nodes and edges into an interaction relationship map of liver cancer genes and space to support efficient query and visual display.

[0110] Stratify and map the liver cancer gene signature data by time slices and store them in the interaction relationship map to form a liver cancer gene database; each time slice corresponds to an independent data layer, supporting query by time point and visual display.

[0111] When new liver cancer gene signature data is added, only the affected nodes and their associated information are updated. If a node appears in the new time slice, a new node information unit is generated and associated with the interaction relationship map; if a node disappears, it is marked as a "historical node" and its evolutionary path is retained.

[0112] In summary, by integrating spatial attributes, representative features, data units, and gene expression correlation weight matrices, this method constructs a liver cancer gene database, which can comprehensively characterize the spatial interaction network of cells, the dynamic evolution of functional modules, and the gene expression correlation characteristics in liver cancer tissues. By adopting a graph database storage architecture, efficient query and visual display of data are realized, while dynamic update and historical data traceability are supported, providing a high-quality data support and analysis platform for liver cancer mechanism research.

[0113] In this embodiment, the method for constructing the liver cancer gene database may further include the following steps:

[0114] Based on the gene expression correlation weight matrix, select gene pairs with weight values higher than a preset weight threshold (such as 0.8) as high-weight gene pairs, and use the high-weight gene pairs as candidate driver genes.

[0115] Construct a co-expression network with candidate driver genes as nodes and gene expression correlation weights as edges, where the higher the weight value, the stronger the co-expression relationship between genes.

[0116] Use network topology metrics (such as node degree, betweenness centrality) to identify key nodes in the co-expression network. If the degree value or betweenness centrality of a gene node is higher than a preset value, it is marked as a key driver gene.

[0117] Perform functional annotation on key driver genes. Based on the GO and KEGG databases, use the hypergeometric distribution method to test and label the biological functions and pathways of key driver genes.

[0118] Extract potential therapeutic targets from the biological functions and pathways of key driver genes, and verify the druggability of the targets in combination with a drug target database (such as DrugBank). If there are existing known drugs or clinical trial data for the target, it is marked as a recommended target.

[0119] Create independent data units for each recommended target in the liver cancer gene database, record the associated key driver genes, biological functions, pathway information, and druggability verification results, and associate them with the original gene expression data, spatial coordinates, and hypergraph structure to form a complete target information chain.

[0120] Among them, the hypergeometric distribution method is a statistical method used to evaluate the probability of a certain event occurring when randomly selecting a specific subset from a finite population. In functional enrichment analysis, its core idea is to compare the key driver gene set with the functional categories in the GO or KEGG database, and by calculating the probability (p-value) of the gene set being enriched in a specific functional category, determine whether it is significantly higher than the random expectation value. If the p-value is less than the preset threshold, it indicates that the functional category is significantly enriched, thereby revealing the biological functions and pathways of the key driver genes.

[0121] In summary, this method can accurately screen potential therapeutic targets for liver cancer by screening high-weight gene pairs based on the gene expression correlation weight matrix, constructing a co-expression network and identifying key driver genes, combining functional annotation and drug target database verification, and mapping the target information back to the liver cancer gene database to form a complete target information chain, providing technical support for the development of targeted drugs for liver cancer.

[0122] In this embodiment, the method for constructing the liver cancer gene database may further include the following steps:

[0123] Based on the gene expression data in the functional module hyperedge, identify immune cell-related genes (such as CD8A, FOXP3), and classify cells using known immune cell markers (such as T cells, B cells, macrophages) to generate an immune cell distribution map;

[0124] Based on the spatial neighborhood hyperedge, analyze the spatial distribution characteristics of immune cells in tumor tissues. If the immune cell density in a certain area is higher than the preset density threshold, it is marked as an "immune cell infiltration hot spot"; if the immune cell density in a certain area is lower than the immune cell density in the adjacent area, it is marked as an "immune rejection area";

[0125] Associate the immune cell distribution map with the functional module hyperedge. If immune cell-related genes are significantly enriched within a certain functional module hyperedge, mark this module as an "immune-related functional module" and record its spatial distribution range;

[0126] Create independent data units for each immune cell infiltration hot spot, immune rejection area, and immune-related functional module, record their immune cell types, densities, functional module information, and spatial distribution ranges, and associate them with the original gene expression data, spatial coordinates, and hypergraph structure to form a complete immune microenvironment feature chain, which is supplemented to the liver cancer gene database.

[0127] In summary, by extracting immune cell distribution features based on functional module hyperedges and spatial neighborhood hyperedges, the present method can identify immune cell infiltration hotspots and immune rejection regions in tumor tissues, and combine functional module hyperedges to label immune-related functional modules, generating data units recording immune cell types, densities, functional module information, and spatial distribution ranges, providing data support for revealing the heterogeneity of the tumor immune microenvironment and optimizing immune treatment strategies.

[0128] As Figure 3 shown, in the second aspect of the present invention, a construction system 8 of a liver cancer gene database is disclosed. The construction system of the liver cancer gene database includes a memory 60 and a processor 80. A construction method program of the liver cancer gene database is stored in the memory 60. When the construction method program of the liver cancer gene database is executed by the processor 80, the steps of any one of the construction methods of the liver cancer gene database are implemented.

[0129] In the third aspect of the present invention, a computer-readable storage medium is disclosed. The computer-readable storage medium includes a construction method program of the liver cancer gene database. When the construction method program of the liver cancer gene database is executed by a processor, the steps of any one of the construction methods of the liver cancer gene database are implemented.

[0130] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other ways. The device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined, or can be integrated into another system, or some features can be ignored, or not executed. In addition, the coupling, direct coupling, or communication connection between the components shown or discussed with each other can be through some interfaces. The indirect coupling or communication connection of devices or units can be electrical, mechanical, or other forms.

[0131] The units described above as separate components may or may not be physically separated. The components shown as units may or may not be physical units; they can be located in one place or distributed to multiple network units; some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0132] In addition, in each embodiment of the present invention, each functional unit can be all integrated in a processing unit, or each unit can be separately used as a unit, or two or more units can be integrated in one unit; the above integrated unit can be implemented in the form of hardware, or in the form of a hardware plus a software functional unit.

[0133] Those of ordinary skill in the art can understand that all or part of the steps to implement the above method embodiments can be completed by hardware related to program instructions. The foregoing program can be stored in a computer-readable storage medium. When the program is executed, it performs the steps including the above method embodiments; and the foregoing storage medium includes: various media such as removable storage devices, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs that can store program codes.

[0134] Alternatively, if the above integrated unit of the present invention is implemented in the form of a software function module and sold or used as an independent product, it can also be stored in a computer-readable storage medium. Based on such an understanding, the technical solution of the embodiments of the present invention, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods of the various embodiments of the present invention. And the foregoing storage medium includes: various media such as removable storage devices, ROM, RAM, magnetic disks, or optical discs that can store program codes.

[0135] The above is only the specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present invention can easily think of changes or substitutions, which should all be covered by the protection scope of the present invention.

Claims

1. A method for constructing a liver cancer gene database, characterized in that It includes the following steps: Obtain the spatial transcriptome data of liver cancer tissues, including gene expression values, cell spatial coordinates, and time series sampling data; Construct a spatial neighborhood relationship based on cell spatial coordinates, define the interconnected cell groups as spatial neighborhood hyperedges, and obtain the spatial attributes of the spatial neighborhood hyperedges; Label functional modules according to gene expression values, associate cell nodes within the same functional module as functional module hyperedges, and obtain the representative features of the functional module hyperedges; Construct a spatio-temporal dynamic hypergraph structure based on the spatial neighborhood hyperedges and functional module hyperedges, and obtain the data units of the spatio-temporal dynamic hypergraph structure; Perform local aggregation on the node features of the spatio-temporal dynamic hypergraph structure to obtain the gene expression correlation weight matrix between the target node and the domain nodes; Construct a liver cancer gene database according to the spatial attributes, representative features, data units, and gene expression correlation weight matrix; Among them, constructing a spatial neighborhood relationship based on cell spatial coordinates, defining the interconnected cell groups as spatial neighborhood hyperedges, and obtaining the spatial attributes of the spatial neighborhood hyperedges, specifically: Based on the cell spatial coordinates of the cells, connect each cell into a triangle by the Delaunay triangulation method to generate a triangular grid graph; Traverse all the triangles in the triangular grid graph, label the two cells sharing the same edge as a spatial neighborhood relationship, and construct an adjacency matrix between the cells; Based on the adjacency matrix, obtain all the interconnected cell groups. If a group of cells is interconnected through neighborhood relationships, then define this group of cells as a spatial neighborhood hyperedge, and record the unique identifiers of all cells within the spatial neighborhood hyperedge; Label spatial attributes for each spatial neighborhood hyperedge, including hyperedge center coordinates, coverage area, cell number, and hyperedge type; Among them, labeling functional modules according to gene expression values, associating cell nodes within the same functional module as functional module hyperedges, and obtaining the representative features of the functional module hyperedges, specifically: Calculate the gene expression similarity between cells based on the gene expression values to generate a symmetric similarity matrix; Based on the similarity matrix, use the hierarchical clustering method to recursively merge the most similar cells or cell clusters, gradually generate a clustering hierarchy until all cells are grouped into a single cluster, and obtain a cell clustering dendrogram; Calculate the silhouette coefficient of each clustering module in the cell clustering dendrogram; if the silhouette coefficient of a certain clustering module is greater than the preset coefficient threshold, then divide this module into an independent co-expression gene module, and record the unique identifiers of all cells within the module; Perform functional enrichment analysis on each co-expression gene module. If the enrichment significance of a certain co-expression gene module is lower than the preset significance threshold, then label it as a functional module; Associate cell nodes within the same functional module as functional module hyperedges, record the unique identifiers of cells within the functional module hyperedges and the functional module type, and at the same time use the central expression profile of the functional module hyperedge as its representative feature.

2. The construction method of a liver cancer gene database according to claim 1, wherein, It also includes: If the number of cells within a certain spatial neighborhood hyperedge is less than the preset value, then determine it as an isolated hyperedge and eliminate it; if the spatial distribution density of cells within a certain spatial neighborhood hyperedge is lower than the preset density threshold, then determine it as a sparse hyperedge; For sparse hyperedges, the Euclidean distance between them and adjacent hyperedges is recalculated. If the Euclidean distance is less than the preset merging threshold, the sparse hyperedge and the adjacent hyperedge are merged into a new spatial neighborhood hyperedge, and the unique identifier list of the cells in the hyperedge is updated.

3. The construction method of a liver cancer gene database according to claim 1, characterized in that Based on the spatial neighborhood hyperedge and the functional module hyperedge, a spatiotemporal dynamic hypergraph structure is constructed, and data units of the spatiotemporal dynamic hypergraph structure are obtained, specifically: The time series sampling data is divided into multiple time slices based on the time series sampling points, each slice contains the gene expression value, cell space coordinates and hyperedge connection status at the current time point; In the first time slice, a hypergraph structure is constructed based on the spatial neighborhood hyperedges and the functional module hyperedges, and a connection relationship between the hypergraph nodes and the spatial neighborhood hyperedges and the functional module hyperedges is recorded; For subsequent time slices, all cell nodes are traversed. If their spatial coordinates or gene expression values ​​change, their hyperedges are recalculated and the hyperedge connection status is updated; If a hyperedge remains unchanged in consecutive time slices, it is marked as a "persistent hyperedge"; if a hyperedge disappears in a certain time slice, it is marked as a "disappearing hyperedge"; if a hyperedge appears for the first time in a certain time slice, it is marked as a "newly added hyperedge"; If a persistent hyperedge shows similar gene expression patterns or spatial distribution characteristics in several time slices, it is marked as a "spatiotemporally stable hyperedge" and its evolution path is stored; If the disappeared hyperedge reappears in the subsequent time slice, it will be restored to a "persistent hyperedge"; if the newly added hyperedge does not reappear in the subsequent time slice, it will be marked as a "temporary hyperedge" and removed; The hypergraph structure of each time slice is stored as an independent data unit, which contains the hyperedge connection status, cell node attributes and hyperedge evolution events.

4. The construction method of a liver cancer gene database according to claim 1, characterized in that The node features of the spatiotemporal dynamic hypergraph structure are locally aggregated to obtain the gene expression correlation weight matrix between the target node and the domain node, specifically: Extract the gene expression characteristics of the cell nodes in each time slice from the spatiotemporal dynamic hypergraph structure, including gene expression values, spatial coordinates and hyperedge information; For each cell node, its local neighborhood range is defined based on its spatial neighborhood hyperedge and functional module hyperedge. If the node belongs to multiple hyperedges at the same time, its neighborhood range is extended to the cell nodes in all related hyperedges. In the local neighborhood, the gene expression similarity between the target node and the neighboring nodes is calculated; if the similarity value is greater than the preset similarity threshold, the target node and the neighboring node are marked as a high-similarity node pair; If it is a high similarity node pair, a first weight value is assigned; otherwise, a second weight value or a zero value is assigned; and a gene expression correlation weight matrix is ​​obtained; wherein the first weight value is greater than the second weight value.

5. The construction method of a liver cancer gene database according to claim 1, wherein, A liver cancer gene database is constructed according to the spatial attributes, representative features, data units and gene expression correlation weight matrix, specifically: The spatial attributes of the spatial neighborhood hyperedges, the representative features of the functional module hyperedges, the data units of the spatiotemporal dynamic hypergraph structure, and the gene expression correlation weight matrix are integrated to form liver cancer gene feature data; Using a graph database as the storage architecture, with hypergraph nodes as entities and spatial neighborhood relationships, temporal evolution paths, and gene expression correlations as edges, construct a map of the interaction relationship between liver cancer genes and space; Store the liver cancer gene feature data in a hierarchical manner by time slices in the interaction relationship map to form a liver cancer gene database; where each time slice corresponds to an independent data layer, supporting query and visualization display by time point; When new liver cancer gene feature data is added, only update the affected nodes and their associated information. If a node appears in the new time slice, generate a new node information unit and associate it with the interaction relationship map; if a node disappears, mark it as a "historical node" and retain its evolution path.

6. A construction system for a liver cancer gene database, characterized in that, The construction system of the liver cancer gene database includes a memory and a processor. The memory stores a program for the construction method of the liver cancer gene database. When the program for the construction method of the liver cancer gene database is executed by the processor, the steps of the construction method of the liver cancer gene database as described in any one of claims 1 to 5 are implemented.

7. A computer-readable storage medium, characterized in that, The computer-readable storage medium includes a program for the construction method of the liver cancer gene database. When the program for the construction method of the liver cancer gene database is executed by the processor, the steps of the construction method of the liver cancer gene database as described in any one of claims 1 to 5 are implemented.

Images