Gene cluster and protein complex detection method and system based on multi-hypergraph coupling

Abstract

The application belongs to the technical field of life science and computer science, and provides a gene cluster and protein complex detection method and system based on multi-supergroup coupling, which comprises the following steps: obtaining gene and protein data of a species from a public biological database, constructing a gene supergroup and a protein supergroup based on the relationship between the gene data and the protein data of the species; constructing an inter-layer edge relationship between the gene supergroup and the protein supergroup; constructing a multi-supergroup coupling random block model with shared potential functional modules; obtaining a community membership matrix of proteins and genes in different relationship supergroups through joint inference, and then identifying the gene cluster and the protein complex. Through the method, the accuracy of gene cluster and protein complex detection is improved.

Description

Technical Field

[0001] This invention relates to the fields of life science technology and computer science, specifically to a method and system for detecting gene clusters and protein complexes based on multi-hypergraph coupling. Background Technology

[0002] With the development of high-throughput sequencing and protein-protein interaction detection technologies, a large amount of heterogeneous and multi-source biological network data has been generated in biological systems, such as gene co-expression data, protein-protein interaction data, and gene-protein association data. These data reflect the functional relationships between biomolecules at different levels, but their forms are complex, often involving higher-order interactions between three or more biomolecules. Traditional graph structures based on simple binary relationships are insufficient to fully characterize these higher-order associations, thus limiting our understanding of the overall structure of biological systems. In actual biological systems, multiple genes participate in biological processes within the same biological pathway or regulatory unit, and multiple proteins function synergistically in the same pathway or molecular complex. These phenomena essentially represent higher-order relationships involving multiple nodes.

[0003] In biological systems, gene clusters and protein complexes are considered key functional units that perform specific biological functions; their essence can be viewed as functional modules with tightly connected characteristics in a hypergraph. Detecting the structure of potential modules within the hypergraph helps to reveal the intrinsic organizational patterns of biological systems.

[0004] While existing technologies have made some progress in clustering or module detection methods for hypergraphs, most only analyze single types of hypergraphs, neglecting the intrinsic connections between different relational perspectives within biological systems. For example, Chinese invention patent application CN113470738A, "A Method and System for Identifying Overlapping Protein Complexes Based on Fuzzy Clustering and Gene Ontology Semantic Similarity," identifies protein complexes by comprehensively considering the interaction relationships between proteins in a protein-protein interaction network and the gene ontology semantic similarity between proteins. Furthermore, in real-world biological scenarios, gene-level regulatory relationships and protein-level interaction relationships are closely related and mutually influential; for instance, gene co-expression relationships and the interaction relationships between their encoded proteins are functionally consistent.

[0005] Therefore, biological systems contain multiple interconnected high-order network structures, and there is an urgent need for a method that can jointly model multiple biological hypergraphs and characterize the coupling relationships between different hypergraphs, so as to achieve more comprehensive and accurate detection of gene clusters and protein complexes. Summary of the Invention

[0006] The technical problem to be solved by this invention is how to improve the accuracy and stability of gene cluster and protein complex detection by coupling gene supermap and protein supermap modeling.

[0007] The present invention solves the above-mentioned technical problems through the following technical means:

[0008] This invention provides a method for detecting gene clusters and protein complexes based on multi-hypergraph coupling, comprising the following steps: S1. Obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph ; S2. Based on the known coding correspondence between proteins and genes in a species, construct inter-layer edge relationships between the gene hypergraph and the protein hypergraph. ; S3. Jointly model the gene supergraph and the protein supergraph to construct a multi-hypergraph coupled random block model with shared potential functional modules; S4. By performing joint inference on the multi-hypergraph coupled random block model, the community membership matrices of proteins and genes in different relational hypergraphs are obtained. and ; S5, Based on Community Membership Matrix and It identifies gene clusters and protein complexes.

[0009] Further, the hypergraphs described in step S1 are all represented as H={V, E}, where the point set V represents biomolecule nodes, the point set of the gene hypergraph is composed of gene nodes, and the point set of the protein hypergraph is composed of protein nodes; the hyperedge set E represents the high-order association relationships formed by the joint participation of multiple biomolecules; each hyperedge... Corresponding to a hyperedge weight The hyperedge weight The strength, reliability, or frequency of occurrence of the biomolecular relationships represented by the hyperedge constitutes the hyperedge weight set. .

[0010] Furthermore, the inter-layer edge relationship described in step S2 This describes the mapping relationship between genes and their corresponding proteins; all inter-layer edges together constitute the inter-layer edge set, used to achieve information alignment and cross-layer transfer between the gene layer and the protein layer in the model, thereby establishing the coupling relationship between different hypergraphs. The inter-layer edge relationship is denoted as... ,in, Represents a set of gene points. This represents a protein point set.

[0011] Further, step S3 includes the following steps: S31. Define the community membership matrix for genes and proteins. ,and ; S32. Introduce relationship-specific in-layer affinity matrices for the gene hypergraph and protein hypergraph, respectively. and This is used to describe the interactive relationships between communities; S33. Introduce interlayer correlation parameters for gene supermaps and protein supermaps. This is used to describe the coupling relationship between communities under different relationships; S34. Based on the introduced parameters and Poisson distribution, a multi-hypergraph coupled random block model is constructed, as follows:

[0012] in, , Pois This represents the Poisson distribution.

[0013] Further, step S4 includes the following steps: S41. Construct the log-likelihood function of a multi-hypergraph coupled random block model; S42. Based on the log-likelihood function, jointly infer the parameters of the multi-hypergraph coupled random block model until the likelihood function converges.

[0014] Furthermore, the log-likelihood function described in step S41 is as follows:

[0015] in, .

[0016] Further, step S42 includes the following steps: S421. By taking the partial derivative of the likelihood function, we obtain the following equation:

[0017]

[0018]

[0019] in, , indicating a protein layer or gene layer, i, j This represents any node in the corresponding hypergraph. k, q, c Indicates the communities that may exist in the hypergraph; 、 and The variational distribution is represented by the following equation:

[0020]

[0021]

[0022]

[0023] S422. Based on the calculation results of step S421, the Expectation-Maximization (EM) algorithm is used to iteratively update the model parameters. This allows us to obtain the community membership matrix of genes and proteins. ,and .

[0024] Further, step S5 includes the following steps: S51. Based on the update results of the community membership matrix, determine the community, i.e., functional module membership relationship of gene nodes and protein nodes respectively; for each gene node or protein node, read the membership value of each potential functional module in the corresponding row of the community membership matrix, wherein the membership value is used to characterize the degree of membership of the node to the corresponding potential functional module; assign the gene node or protein node to the potential functional module with the largest membership value, and gather the nodes assigned to the same potential functional module to form a candidate functional module composed of multiple gene nodes or multiple protein nodes. S52. Construct screening rules for candidate functional modules, including: when the set of gene layer nodes and the set of protein layer nodes in the candidate functional module show at least 70% correspondence through interlayer mapping, output the candidate functional module as a gene cluster or protein complex. S53. Traverse all candidate functional modules, output all candidate functional modules that meet the screening rules, and complete the detection of gene clusters and protein complexes in the biological system.

[0025] This invention also provides a gene cluster and protein complex detection system based on multi-hypergraph coupling. The system executes the above-described method during operation and includes the following modules: The hypergraph construction module is used to obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph ; The interlayer edge relationship construction module is used to construct interlayer edge relationships between gene supergraphs and protein supergraphs based on the known coding correspondence between proteins and genes in a species. ; The coupling model construction module is used to jointly model gene hypergraphs and protein hypergraphs, and construct multi-hypergraph coupled random block models with shared potential functional modules. The joint inference module is used to obtain the community membership matrices of proteins and genes in different relational hypergraphs by performing joint inference on a multi-hypergraph coupled random block model. and ; Output module, used for community membership matrix and It identifies gene clusters and protein complexes.

[0026] In addition, the present invention provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method.

[0027] The advantages of this invention are: (1) The present invention adopts a hypergraph modeling method, which breaks through the limitations of the traditional binary relation graph model and can represent the high-order biological relationships involving multiple genes or multiple proteins, thereby more realistically depicting the structural features of multi-molecule synergistic effects in biological systems and improving the modeling and expression capabilities.

[0028] (2) This invention constructs gene supergraphs and protein supergraphs and performs coupled modeling to achieve the synergistic utilization of multi-source biological relationship information, avoids the information loss caused by single relationship network analysis, and effectively improves the stability and reliability of gene cluster and protein complex detection.

[0029] (3) The multi-hypergraph coupling modeling framework of the present invention does not depend on specific biological relationship types and can be flexibly extended to other high-order biological relationship data (such as transcriptional regulation relationship, metabolic response relationship, etc.), and has strong versatility and application promotion value. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the process for detecting gene clusters and protein complexes based on multi-hypergraph coupling, according to an embodiment of the present invention. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0032] Example 1 This embodiment provides a method for detecting gene clusters and protein complexes based on multi-hypergraph coupling, addressing the problem that existing methods for detecting gene clusters and protein complexes often rely on single-relationship networks and neglect the intrinsic coupling between multiple biological relationships. This method constructs gene and protein hypergraphs incorporating multiple higher-order biological relationships, introduces shared latent functional module representations, and jointly models the multi-relationship hypergraphs. By utilizing the natural associations between genes and proteins, information complementarity is achieved, thereby improving the accuracy and stability of gene cluster and protein complex detection. The specific implementation process is as follows: Figure 1 As shown, it includes the following steps: S1. Obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph The hyperedges of the gene and protein hypergraphs can be constructed based on biological pathways, biological function annotations, or molecular synergistic relationships, and are used to characterize higher-order relationships between biomolecules from different biological relationship perspectives.

[0033] The hypergraphs described are all represented as H={V, E}, where the point set V represents biomolecule nodes, with the point set of the gene hypergraph consisting of gene nodes and the point set of the protein hypergraph consisting of protein nodes; the hyperedge set E represents the high-order associations formed by the joint participation of multiple biomolecules; each hyperedge... Corresponding to a hyperedge weight The hyperedge weight The strength, reliability, or frequency of occurrence of the biomolecular relationships represented by the hyperedge constitutes the hyperedge weight set. In the gene hypergraph, each hyperedge represents a group of genes participating in a biological process within the same biological pathway, functional module, or regulatory unit, depicting high-order relationships between genes at the functional regulatory level. Similarly, each hyperedge in the protein hypergraph represents interactions between multiple protein molecules within the same biological pathway or molecular complex, reflecting the cooperative execution relationships of proteins in a biological system. Through these methods, high-order functional structures in biological systems are modeled from both the regulatory and execution levels.

[0034] S2. Based on the known coding correspondence between proteins and genes in a species, construct inter-layer edge relationships between the gene hypergraph and the protein hypergraph. It is used to establish mappings and couplings between different hypergraphs, thereby enabling cross-hypergraph information association and collaborative modeling.

[0035] The interlayer edge relationship This is used to describe the mapping relationship between genes and their corresponding proteins. Specifically, when a gene and a protein have a corresponding relationship, inter-layer edges are constructed between the gene hypergraph and the protein hypergraph to represent the gene and protein. All inter-layer edges together constitute an inter-layer edge set, which is used to realize information alignment and cross-layer transfer between the gene layer and the protein layer in the model, thereby establishing the coupling relationship between different hypergraphs. The inter-layer edge relationship is denoted as... ,in, Represents a set of gene points. This represents a protein point set.

[0036] S3. Jointly model the gene supergraph and protein supergraph to construct a multi-hypergraph coupled random block model with shared potential functional modules. Based on the construction of the gene supergraph, protein supergraph, and their interlayer edges, the multi-relationship biological supergraph is treated as a whole and introduced into the multi-hypergraph random block model for modeling. Shared potential functional modules are introduced for genes and proteins to describe the potential functional attribution of molecules from different relational perspectives. By sharing potential functional modules, gene regulatory information and protein interaction information are fused at the model level. Specific implementation methods include the following steps: S31. Define the community membership matrix for genes and proteins. ,and ; S32. Introduce relationship-specific in-layer affinity matrices for the gene hypergraph and protein hypergraph, respectively. and This is used to describe the interactive relationships between communities; S33. Introduce interlayer correlation parameters for gene supermaps and protein supermaps. This is used to describe the coupling relationship between communities under different relationships; S34. Based on the introduced parameters and Poisson distribution, a multi-hypergraph coupled random block model is constructed, as follows:

[0037] in, , Pois This represents the Poisson distribution.

[0038] S4. By performing joint inference on the multi-hypergraph coupled random block model, the community membership matrices of proteins and genes in different relational hypergraphs are obtained. and The specific implementation method includes the following steps: S41. Construct the log-likelihood function of the multi-hypergraph coupled random block model; as shown in the following equation:

[0039] in, .

[0040] S42. Based on the log-likelihood function, jointly infer the parameters of the multi-hypergraph coupled random block model until the likelihood function converges. The specific implementation includes the following steps: S421. By taking the partial derivative of the likelihood function, we obtain the following equation:

[0041]

[0042]

[0043] in, , indicating a protein layer or gene layer, i, j This represents any node in the corresponding hypergraph. k, q, c Indicates the communities that may exist in the hypergraph; 、 and The variational distribution is represented by the following equation:

[0044]

[0045]

[0046]

[0047] S422. Based on the calculation results of step S421, the Expectation-Maximization (EM) algorithm is used to iteratively update the model parameters. This allows us to obtain the community membership matrix of genes and proteins. ,and .

[0048] S5, Based on Community Membership Matrix and The method identifies gene clusters and protein complexes. Specific implementation methods include the following steps: S51. Based on the update results of the community membership matrix, determine the community, i.e., functional module membership relationship of gene nodes and protein nodes respectively; for each gene node or protein node, read the membership value of each potential functional module in the corresponding row of the community membership matrix, wherein the membership value is used to characterize the degree of membership of the node to the corresponding potential functional module; assign the gene node or protein node to the potential functional module with the largest membership value, and gather the nodes assigned to the same potential functional module to form a candidate functional module composed of multiple gene nodes or multiple protein nodes. S52. Construct screening rules for candidate functional modules, including: when the set of gene layer nodes and the set of protein layer nodes in the candidate functional module show at least 70% correspondence through inter-layer mapping relationship, the candidate functional module is output as a gene cluster or protein complex; wherein, the correspondence ratio is defined as the ratio of "the number of nodes that can find the mapping object in another layer" to "the total number of potential functional module nodes in this layer".

[0049] S53. Traverse all candidate functional modules, output all candidate functional modules that meet the screening rules, and complete the detection of gene clusters and protein complexes in the biological system.

[0050] In this embodiment, the above method is used on a self-built validation dataset, and the HMC protein complex (UniProt IDs: [P02829, P15108, P10591, P25567, P0CE41, P10592, P25491]) and its associated gene cluster (gene IDs: [YPL240C, YMR186W, YAL005C, YLL024C, YNL064C]) are used as validation objects, and precision and recall are used for evaluation. Experimental results show that the method described in this embodiment can effectively integrate cross-domain complementary information in multi-hypergraph structures, thereby achieving high-quality recovery of functional modules. Specifically, at the gene level, the communities detected by the model are completely consistent with the real gene clusters, with both precision and recall reaching 1.0. At the protein level, the model's detection results have a precision of 1.0 and a recall of 0.7142, failing to reach full recall only due to the omission of two proteins (P0CE41 and P25567). Overall, the interlayer edges in the multi-hypergraph characterize the functional association between protein interactions and gene regulation, thus significantly improving the accuracy and reliability of the method described in this embodiment in complex and gene cluster detection tasks.

[0051] Example 2 It should be further noted that, based on the same inventive concept, this embodiment also provides a gene cluster and protein complex detection system based on multi-hypergraph coupling. When the system is running, it executes the method described in Embodiment 1, including the following modules: The hypergraph construction module is used to obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph ; The interlayer edge relationship construction module is used to construct interlayer edge relationships between gene supergraphs and protein supergraphs based on the known coding correspondence between proteins and genes in a species. ; The coupling model construction module is used to jointly model gene supergraphs and protein supergraphs, and construct multi-hypergraph coupled random block models with shared potential functional modules. The joint inference module is used to obtain the community membership matrices of proteins and genes in different relational hypergraphs by performing joint inference on a multi-hypergraph coupled random block model. and ; Output module, used for community membership matrix and It identifies gene clusters and protein complexes.

[0052] Furthermore, this embodiment also proposes a computer-readable storage medium storing a computer program, the computer program including program instructions, and the processor executing the program instructions to implement the gene cluster and protein complex detection method based on multi-hypergraph coupling described in Embodiment 1.

[0053] For the purposes of this specification, "computer storage medium" can be any means capable of containing, storing, communicating, propagating, or transmitting a program for use in or in conjunction with an instruction execution system, apparatus, or device. More specific examples of computer storage media (a non-exhaustive list) include: an electrical connection having one or more wires (electronic device), a portable computer disk drive (magnetic device), random access memory (RAM), read-only memory (ROM), erasable and editable read-only memory (EPROM or flash memory), fiber optic devices, and portable optical disc read-only memory (CDROM). Furthermore, a computer storage medium can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, interpreting, or otherwise processing as necessary, and then stored in a computer memory.

[0054] It should be understood that various parts of the present invention can be implemented in hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented in software or firmware stored in memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, it can be implemented using any one or a combination of the following techniques known in the art: discrete logic circuits having logic gates for implementing logical functions on data signals, application-specific integrated circuits (ASICs) having suitable combinational logic gates, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0055] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for detecting gene clusters and protein complexes based on multi-hypergraph coupling, characterized in that, Includes the following steps: S1. Obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph ; S2. Based on the known coding correspondence between proteins and genes in a species, construct inter-layer edge relationships between the gene hypergraph and the protein hypergraph. ; S3. Jointly model the gene supergraph and the protein supergraph to construct a multi-hypergraph coupled random block model with shared potential functional modules; S4. By performing joint inference on the multi-hypergraph coupled random block model, the community membership matrices of proteins and genes in different relational hypergraphs are obtained. and ; S5, Based on Community Membership Matrix and It identifies gene clusters and protein complexes.

2. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 1, characterized in that, The hypergraphs described in step S1 are all represented as H={V, E}, where the point set V represents biomolecule nodes, with the point set of the gene hypergraph consisting of gene nodes and the point set of the protein hypergraph consisting of protein nodes; the hyperedge set E represents the high-order association relationships formed by the joint participation of multiple biomolecules; each hyperedge... Corresponding to a hyperedge weight The hyperedge weight The strength, reliability, or frequency of occurrence of the biomolecular relationships represented by the hyperedge constitutes the hyperedge weight set. .

3. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 2, characterized in that, Step S2 describes the interlayer edge relationship Used to describe the mapping relationship between genes and their corresponding proteins; All inter-layer edges together constitute the inter-layer edge set, which is used to realize information alignment and cross-layer transfer between the gene layer and the protein layer in the model, thereby establishing the coupling relationship between different hypergraphs. The inter-layer edge relationship is denoted as . ,in, Represents a set of gene points. This represents a protein point set.

4. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 1, characterized in that, Step S3 includes the following steps: S31. Define the community membership matrix for genes and proteins. ,and ; S32. Introduce relationship-specific in-layer affinity matrices for the gene hypergraph and protein hypergraph, respectively. and This is used to describe the interactive relationships between communities; S33. Introduce interlayer correlation parameters for gene supermaps and protein supermaps. This is used to describe the coupling relationship between communities under different relationships; S34. Based on the introduced parameters and Poisson distribution, a multi-hypergraph coupled random block model is constructed, as follows: in, , Pois This represents the Poisson distribution.

5. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 4, characterized in that, Step S4 includes the following steps: S41. Construct the log-likelihood function of a multi-hypergraph coupled random block model; S42. Based on the log-likelihood function, jointly infer the parameters of the multi-hypergraph coupled random block model until the likelihood function converges.

6. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 5, characterized in that, The log-likelihood function mentioned in step S41 is as follows: in, .

7. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 6, characterized in that, Step S42 includes the following steps: S421. By taking the partial derivative of the likelihood function, we obtain the following equation: in, , indicating a protein layer or gene layer, i, j This represents any node in the corresponding hypergraph. k、 q, c Indicates the communities that may exist in the hypergraph; 、 and The variational distribution is represented by the following equation: S422. Based on the calculation results of step S421, the Expectation-Maximization (EM) algorithm is used to iteratively update the model parameters. This allows us to obtain the community membership matrix of genes and proteins. ,and .

8. The method for detecting gene clusters and protein complexes based on multi-hypergraph coupling according to claim 1, characterized in that, Step S5 includes the following steps: S51. Based on the updated results of the community membership matrix, determine the community, i.e., functional module membership, of gene nodes and protein nodes respectively; for each gene node or protein node, read the membership value of each potential functional module in the corresponding row of the community membership matrix, wherein the membership value is used to characterize the degree of membership of the node to the corresponding potential functional module; assign the gene node or protein node to the potential functional module with the largest membership value, and gather the nodes assigned to the same potential functional module to form a candidate functional module composed of multiple gene nodes or multiple protein nodes. S52. Construct screening rules for candidate functional modules, including: when the set of gene layer nodes and the set of protein layer nodes in the candidate functional module show at least 70% correspondence through interlayer mapping, output the candidate functional module as a gene cluster or protein complex. S53. Traverse all candidate functional modules, output all candidate functional modules that meet the screening rules, and complete the detection of gene clusters and protein complexes in the biological system.

9. A gene cluster and protein complex detection system based on multi-hypergraph coupling, characterized in that, Includes the following modules: The hypergraph construction module is used to obtain gene and protein data of species from publicly available biological databases, and construct gene hypergraphs based on the relationships between the gene and protein data of the species. and protein hypergraph ; The interlayer edge relationship construction module is used to construct interlayer edge relationships between gene supergraphs and protein supergraphs based on the known coding correspondence between proteins and genes in a species. ; The coupling model construction module is used to jointly model gene supergraphs and protein supergraphs, and construct multi-hypergraph coupled random block models with shared potential functional modules. The joint inference module is used to obtain the community membership matrices of proteins and genes in different relational hypergraphs by performing joint inference on a multi-hypergraph coupled random block model. and ; Output module, used for community membership matrix and It identifies gene clusters and protein complexes.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the method as described in any one of claims 1-8.

Images