Drug and cancer relationship analysis method and device based on graph neural network

Abstract

This invention relates to the fields of intelligent decision-making and digital healthcare, disclosing a method, apparatus, electronic device, and storage medium for analyzing the relationship between drugs and cancer based on graph neural networks. The method includes: querying multi-omics data and gene regulatory networks of cancer information; querying molecular maps of drug information; determining cancer vectors of cancer information using a graph neural network model; determining drug vectors of drug information using a graph neural network model; constructing a cancer-drug bipartite graph of cancer information and drug information; calculating a first loss value of the graph neural network model; calculating a second loss value between cancer information and drug information; training the graph neural network model using the first and second loss values ​​to obtain a trained graph neural network model; and using the trained graph neural network model to identify the cancer-drug relationship between cancer information and drug information. This invention can improve the effectiveness of drug-cancer relationship analysis.

Description

Technical Field

[0001] This invention relates to the fields of intelligent decision-making and digital healthcare, and in particular to a method and apparatus for analyzing the relationship between drugs and cancer based on graph neural networks. Background Technology

[0002] The analysis of the relationship between drugs and cancer based on graph neural networks refers to the process of using graph neural network models to output the relationship between drugs and cancer, in order to predict the response of different tumor cells to drugs.

[0003] Currently, with the rise of machine learning technology, the digital healthcare field can support functions such as disease-aided diagnosis, health management, and remote consultation. For example, in cancer treatment, neural network models can predict the response of different tumor cells to drugs. However, most existing technologies use textual data or structural information of cancer and drug data as input data for neural network models. Due to the lack of implicit correlation information in image structures, this analysis method is insufficient in extracting the implicit information about the relationship between drugs and cancer. Secondly, most existing technologies train neural network models using historical data of cancer and drugs, lacking training to utilize the model to discover novel relationships between cancer and drug data. Therefore, the analysis of the relationship between drugs and cancer is ineffective. Summary of the Invention

[0004] This invention provides a method and apparatus for analyzing the relationship between drugs and cancer based on graph neural networks. Its main purpose is to obtain new cancer vectors and drug vectors by updating historical data vectors, and to use the graph structure of cancer information and drug information as input data for graph neural network models, thereby improving the analysis effect of the relationship between drugs and cancer.

[0005] To achieve the above objectives, this invention provides a method for analyzing the relationship between drugs and cancer based on graph neural networks, comprising:

[0006] Obtain cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, use a graph neural network model to determine the cancer vector of the cancer information, and use the graph neural network model to determine the drug vector of the drug information.

[0007] Construct a cancer-drug bipartite graph between the cancer information and the drug information, and use the cancer vector, the drug vector, and the cancer-drug bipartite graph to calculate the first loss value of the graph neural network model;

[0008] Based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, calculate a second loss value between the cancer information and the drug information;

[0009] The graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model. The trained graph neural network model is then used to identify the cancer-drug relationship between the cancer information and the drug information.

[0010] Optionally, the multi-omics data and gene regulatory network used to query the cancer information include:

[0011] Identify the target of the cancer information query;

[0012] Based on the query objective, query the multi-omics data of the cancer information;

[0013] Based on the multi-omics data, a prior distribution of the regulatory network for the cancer information is constructed;

[0014] A gene regulation network model for cancer information is constructed using the prior distribution of the regulatory network, and the gene regulation network model is used as the gene regulation network.

[0015] Optionally, the process of querying the molecular map of the drug information includes:

[0016] Determine the name of the drug in the drug information;

[0017] To find the drug components of the drug name;

[0018] Extract the molecular topological structure of the drug's constituent molecules;

[0019] The molecular topology is used to determine the molecular map of the drug information.

[0020] Optionally, constructing the cancer-drug bipartite graph between the cancer information and the drug information includes:

[0021] Determine the cancer vertex of the cancer information and the drug vertex of the drug information;

[0022] Query the historical relationship between the cancer vertex and the drug vertex;

[0023] Based on the historical relationships, construct the relationship edges between the cancer vertex and the drug vertex;

[0024] Based on the relation edges, the cancer vertex, and the drug vertex, a cancer-drug bipartite graph is determined between the cancer information and the drug information.

[0025] Optionally, calculating the first loss value of the graph neural network model using the cancer vector, the drug vector, and the cancer-drug bipartite graph includes:

[0026] Calculate the dot product between the cancer vector and the drug vector;

[0027] Calculate the activation probability value of the vector inner product;

[0028] In the cancer-drug bipartite graph, query the true relationship between the cancer vector and the drug vector;

[0029] Determine the probability of the true relationship.

[0030] Calculate the cross-entropy loss value between the true relation probability and the activation probability value, and use the cross-entropy loss value as the first loss value of the graph neural network model.

[0031] Optionally, calculating the second loss value between the cancer information and the drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph includes:

[0032] The cancer-drug bipartite graph is encoded into a bipartite graph vector.

[0033] The initial node vectors of the cancer-drug bipartite graph are constructed using the cancer vector and the drug vector;

[0034] The node initial vector is updated to obtain the updated node vector;

[0035] The structure of the cancer-drug bipartite graph is scrambled to obtain a scrambled bipartite graph;

[0036] Calculate the shuffled node vector of the shuffled bipartite graph;

[0037] Calculate the first vector distance between the shuffled node vector and the bipartite graph vector, and calculate the second vector distance between the updated node vector and the bipartite graph vector;

[0038] The second loss value is calculated using the following formula based on the first vector distance and the second vector distance:

[0039]

[0040] Where L2 represents the second loss value, D1 represents the first vector distance, that is, the distance between the vector of the i-th shuffled node and the vector of the i-th node in the bipartite graph vector, D2 represents the second vector distance, that is, the distance between the vector of the i-th updated node and the vector of the i-th node in the bipartite graph vector, k represents the number of nodes in the bipartite graph, and i represents the node index in the bipartite graph.

[0041] Optionally, the step of training the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model includes:

[0042] The graph neural network model is backpropagated using the first loss value and the second loss value to obtain the backpropagation model.

[0043] Calculate the output loss value of the backpropagation model;

[0044] When the output loss value is not less than the preset loss value, return to the above steps of using the first loss value and the second loss value to perform backpropagation processing on the graph neural network model to obtain the backpropagation model;

[0045] When the output loss value is less than the preset loss value, the trained graph neural network model is obtained.

[0046] To address the aforementioned problems, the present invention also provides a device for analyzing the relationship between drugs and cancer based on graph neural networks, the device comprising:

[0047] The vector determination module is used to acquire cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, the module uses a graph neural network model to determine the cancer vector of the cancer information and the graph neural network model to determine the drug vector of the drug information.

[0048] The first loss calculation module is used to construct a cancer-drug bipartite graph between the cancer information and the drug information, and to calculate the first loss value of the graph neural network model using the cancer vector, the drug vector and the cancer-drug bipartite graph.

[0049] The second loss calculation module is used to calculate a second loss value between the cancer information and the drug information based on the cancer vector, the drug vector and the cancer-drug bipartite graph;

[0050] The relationship recognition module is used to train the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model, and to use the trained graph neural network model to identify the cancer-drug relationship between the cancer information and the drug information.

[0051] To address the above problems, the present invention also provides an electronic device, the electronic device comprising:

[0052] At least one processor; and,

[0053] A memory communicatively connected to the at least one processor; wherein,

[0054] The memory stores a computer program that can be executed by the at least one processor to implement the above-described method for analyzing the relationship between drugs and cancer based on graph neural networks.

[0055] To address the aforementioned problems, the present invention also provides a computer-readable storage medium storing at least one computer program, which is executed by a processor in an electronic device to implement the aforementioned method for analyzing the relationship between drugs and cancer based on graph neural networks.

[0056] As can be seen, this embodiment of the invention first queries the multi-omics data and gene regulatory network of the cancer information to fuse gene expression and cancer mutation information from different omics based on the gene regulatory network. Further, this embodiment queries the molecular map of the drug information to integrate cancer gene expression information and drug molecule information, ensuring subsequent prediction of the cancer-drug response relationship. Further, this embodiment uses a graph neural network model based on the multi-omics data and the gene regulatory network to determine the cancer vector of the cancer information, to fuse gene expression and mutation information from different omics based on the gene regulatory network. Further, this embodiment uses the graph neural network model to determine the drug vector of the drug information to obtain the feature vector of the drug information. Secondly, this embodiment constructs a cancer-drug bipartite graph between the cancer information and the drug information to construct a graph of the historical correlation between the cancer information and the drug information, allowing the subsequent graph neural network model to learn the historical relationships of the graph. Further, this embodiment... This embodiment calculates a first loss value for the graph neural network model using the cancer vector and the drug vector, and uses this first loss value to train and optimize the graph neural network model, thereby improving the accuracy of the graph neural network model in analyzing the relationship between cancer and drugs. Further, this embodiment calculates a second loss value between the cancer information and the drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, and uses this second loss value to train the graph neural network model using updated cancer vectors and drug vectors. This embodiment trains the graph neural network model using the first and second loss values ​​to obtain a trained graph neural network model, which is then used to construct a graph neural network based on the known drug-cancer interaction relationship, enabling better prediction of cancer-drug relationships. Furthermore, this embodiment uses the trained graph neural network model to identify the cancer-drug relationship between the cancer information and the drug information, and uses the trained graph neural network to output the cancer-drug relationship, thereby improving the accuracy of probability output and better predicting cancer-drug relationships. Therefore, the method, apparatus, electronic device, and computer-readable storage medium for analyzing the relationship between drugs and cancer based on graph neural networks proposed in this invention can obtain new cancer vectors and drug vectors by updating historical data vectors, and use the graph structure of cancer information and drug information as input data for the graph neural network model, thereby improving the analysis effect of the relationship between drugs and cancer. Attached Figure Description

[0057] Figure 1 This is a flowchart illustrating a method for analyzing the relationship between drugs and cancer based on graph neural networks, provided in an embodiment of the present invention.

[0058] Figure 2 This is a schematic diagram of a module of a graph neural network-based drug-cancer relationship analysis device provided in an embodiment of the present invention.

[0059] Figure 3 This is a schematic diagram of the internal structure of an electronic device that implements a method for analyzing the relationship between drugs and cancer based on graph neural networks, according to an embodiment of the present invention.

[0060] The realization of the objective, functional features and advantages of the present invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0061] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.

[0062] This invention provides a method for analyzing the relationship between drugs and cancer based on graph neural networks. The execution entity of this method includes, but is not limited to, at least one of the following electronic devices that can be configured to execute the method provided in this invention: a server, a terminal, etc. In other words, the method can be executed by software or hardware installed on a terminal device or a server device, and the software may be a blockchain platform. The server includes, but is not limited to, a single server, a server cluster, a cloud server, or a cluster of cloud servers. The server can be an independent server or a cloud server providing basic cloud computing services such as cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communication, middleware services, domain name services, security services, content delivery networks (CDNs), and big data and artificial intelligence platforms.

[0063] Reference Figure 1 The diagram shown is a flowchart illustrating a method for analyzing the relationship between drugs and cancer based on graph neural networks, according to an embodiment of the present invention. In this embodiment, the method includes the following steps S1-S4:

[0064] S1. Obtain cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, use a graph neural network model to determine the cancer vector of the cancer information, and use the graph neural network model to determine the drug vector of the drug information.

[0065] This invention utilizes multi-omics data and gene regulatory networks to query cancer information, enabling the fusion of gene expression and cancer mutation information from different omics based on these networks. The multi-omics data refers to data from various omics sources, including data on gene mutations (SNPs), gene expression values, DNA methylation, and other dimensions, for big data processing. The gene regulatory network refers to the network formed by the interactions between genes within a cell (or a specific genome). Among these interactions, it specifically refers to gene-gene interactions resulting from gene regulation, including protein-protein interaction networks and transcription factor-mediated gene regulation networks.

[0066] In one embodiment of the present invention, querying the multi-omics data and gene regulatory network of the cancer information includes: identifying the query target of the cancer information; querying the multi-omics data of the cancer information based on the query target; constructing a prior distribution of the regulatory network of the cancer information according to the multi-omics data; constructing a gene regulatory network model of the cancer information using the prior distribution of the regulatory network, and using the gene regulatory network model as the gene regulatory network.

[0067] Optionally, the multi-omics data query for cancer information based on the query target is achieved by using omics data from a public database, wherein the public database includes TCGA database, GEO database, EBI database, SRA database, CGGA database, PRIDE database, etc.

[0068] For example, the process of constructing the prior distribution of the regulatory network of cancer information based on the multi-omics data and the process of constructing the gene regulatory network model of cancer information using the prior distribution of the regulatory network are as follows: a method of querying DNA sequences and expression profiles in the multi-omics data; using existing literature information to determine the associated transcription factor set and DNA sequence data to identify possible target genes, thereby establishing a prior distribution of the regulatory network; and then establishing a log-linear model and a Bayesian model on the gene expression data to obtain a new regulatory network, regulatory strength, and transcription factor activity.

[0069] Furthermore, embodiments of the present invention utilize molecular maps of the drug information to integrate cancer gene expression information and drug molecule information, ensuring subsequent prediction of the cancer-drug response relationship. The molecular map refers to a map displaying the entire genome structure of an organism. Depending on the research purpose, method, and level of detail in establishing the map, it can take different forms, including genetic linkage maps established using genetic methods, physical maps depicting the distribution of gene molecular positions by distance, nucleotide sequence maps established by determining nucleic acid sequences, and transcription maps marking expressible sequences, etc.

[0070] In one embodiment of the present invention, querying the molecular map of the drug information includes: determining the drug name of the drug information; querying the drug constituent molecules of the drug name; extracting the molecular topology of the drug constituent molecules; and using the molecular topology to determine the molecular map of the drug information.

[0071] Furthermore, in this embodiment of the invention, a cancer vector of the cancer information is determined using a graph neural network model based on the multi-omics data and the gene regulatory network, so as to fuse gene expression and variation information from different omics based on the gene regulatory network.

[0072] In one embodiment of the present invention, the determination of the cancer vector of the cancer information based on the multi-omics data and the gene regulatory network using a graph neural network model is achieved by utilizing the neural network structure in the graph neural network model.

[0073] The neural network structure includes a convolutional neural network structure and a recurrent neural network structure. The convolutional neural network is mainly used for processing image data, and the recurrent neural network is mainly used for processing time-series data.

[0074] Furthermore, in this embodiment of the invention, the drug vector of the drug information is determined by utilizing the graph neural network model, in order to obtain the feature vector of the drug information.

[0075] In one embodiment of the present invention, the principle of determining the drug vector of the drug information using the graph neural network model is similar to the principle of determining the cancer vector of the cancer information using the graph neural network model based on the multi-omics data and the gene regulatory network, and will not be elaborated further here.

[0076] S2. Construct a cancer-drug bipartite graph between the cancer information and the drug information, and use the cancer vector, the drug vector and the cancer-drug bipartite graph to calculate the first loss value of the graph neural network model.

[0077] This invention constructs a cancer-drug bipartite graph between cancer information and drug information to represent the historical relationships between them in a graph-like structure. This allows subsequent graph neural network models to learn these historical relationships. Specifically, the cancer-drug bipartite graph is a graph whose vertex set can be divided into two disjoint subsets, where each edge in the graph is attached to two vertices belonging to either of these disjoint subsets, and the vertices within the two subsets are not adjacent.

[0078] In one embodiment of the present invention, constructing a cancer-drug bipartite graph between the cancer information and the drug information includes: determining the cancer vertices of the cancer information and the drug vertices of the drug information; querying the historical relationships between the cancer vertices and the drug vertices; constructing relationship edges between the cancer vertices and the drug vertices based on the historical relationships; and determining the cancer-drug bipartite graph between the cancer information and the drug information based on the relationship edges, the cancer vertices, and the drug vertices.

[0079] Furthermore, in this embodiment of the invention, a first loss value is calculated for the graph neural network model using the cancer vector and the drug vector, and this first loss value is used to train and optimize the graph neural network model, thereby improving the accuracy of the graph neural network model in analyzing the relationship between cancer and drugs.

[0080] In one embodiment of the present invention, the step of calculating the first loss value of the graph neural network model using the cancer vector, the drug vector, and the cancer-drug bipartite graph includes: calculating the dot product between the cancer vector and the drug vector; calculating the activation probability value of the dot product; querying the true relationship between the cancer vector and the drug vector in the cancer-drug bipartite graph; determining the true relationship probability of the true relationship; calculating the cross-entropy loss value between the true relationship probability and the activation probability value, and using the cross-entropy loss value as the first loss value of the graph neural network model.

[0081] Here, the true relationship refers to the historical relationship. Optionally, the probability of determining the true relationship is achieved by querying the cancer-drug bipartite graph where an edge is 1 and no edge is 0.

[0082] S3. Calculate the second loss value between the cancer information and the drug information based on the cancer vector, the drug vector and the cancer-drug bipartite graph.

[0083] In this embodiment of the invention, a second loss value is calculated between the cancer information and the drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, so as to train the graph neural network model using the updated cancer vector and drug vector.

[0084] In one embodiment of the present invention, the step of calculating a second loss value between cancer information and drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph includes: performing bipartite graph encoding on the cancer-drug bipartite graph to obtain a bipartite graph vector; constructing initial node vectors of the cancer-drug bipartite graph using the cancer vector and the drug vector; updating the initial node vectors to obtain updated node vectors; scrambling the structure of the cancer-drug bipartite graph to obtain a scrambled bipartite graph; calculating the scrambled node vectors of the scrambled bipartite graph; calculating a first vector distance between the scrambled node vectors and the bipartite graph vectors, and calculating a second vector distance between the updated node vectors and the bipartite graph vectors; and calculating the second loss value using the following formula based on the first vector distance and the second vector distance:

[0085]

[0086] Where L2 represents the second loss value, D1 represents the first vector distance, that is, the distance between the vector of the i-th shuffled node and the vector of the i-th node in the bipartite graph vector, D2 represents the second vector distance, that is, the distance between the vector of the i-th updated node and the vector of the i-th node in the bipartite graph vector, k represents the number of nodes in the bipartite graph, and i represents the node index in the bipartite graph.

[0087] S4. The graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model. The trained graph neural network model is then used to identify the cancer-drug relationship between the cancer information and the drug information.

[0088] In this embodiment of the invention, the graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model, which is then used to construct a graph neural network based on the known relationship between drugs and cancer, thereby enabling better prediction of the relationship between cancer and drugs.

[0089] In one embodiment of the present invention, the step of training the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model includes: performing backpropagation processing on the graph neural network model using the first loss value and the second loss value to obtain a backpropagation model; calculating the output loss value of the backpropagation model; when the output loss value is not less than a preset loss value, returning to the step of performing backpropagation processing on the graph neural network model using the first loss value and the second loss value to obtain a backpropagation model; and when the output loss value is less than the preset loss value, obtaining the trained graph neural network model.

[0090] Furthermore, in this embodiment of the invention, the cancer-drug relationship between the cancer information and the drug information is identified by utilizing the trained graph neural network model. This allows the trained graph neural network to output the cancer-drug relationship, thereby improving the accuracy of the probability output and better predicting the cancer-drug relationship.

[0091] As can be seen, this embodiment of the invention first queries the multi-omics data and gene regulatory network of the cancer information to fuse gene expression and cancer mutation information from different omics based on the gene regulatory network. Further, this embodiment queries the molecular map of the drug information to integrate cancer gene expression information and drug molecule information, ensuring subsequent prediction of the cancer-drug response relationship. Further, this embodiment uses a graph neural network model based on the multi-omics data and the gene regulatory network to determine the cancer vector of the cancer information, to fuse gene expression and mutation information from different omics based on the gene regulatory network. Further, this embodiment uses the graph neural network model to determine the drug vector of the drug information to obtain the feature vector of the drug information. Secondly, this embodiment constructs a cancer-drug bipartite graph between the cancer information and the drug information to construct a graph of the historical correlation between the cancer information and the drug information, allowing the subsequent graph neural network model to learn the historical relationships of the graph. Further, this embodiment... This embodiment calculates a first loss value for the graph neural network model using the cancer vector and the drug vector, and uses this first loss value to train and optimize the graph neural network model, thereby improving the accuracy of the graph neural network model in analyzing the relationship between cancer and drugs. Further, this embodiment calculates a second loss value between the cancer information and the drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, and uses this second loss value to train the graph neural network model using updated cancer vectors and drug vectors. This embodiment trains the graph neural network model using the first and second loss values ​​to obtain a trained graph neural network model, which is then used to construct a graph neural network based on the known drug-cancer interaction relationship, enabling better prediction of cancer-drug relationships. Furthermore, this embodiment uses the trained graph neural network model to identify the cancer-drug relationship between the cancer information and the drug information, and uses the trained graph neural network to output the cancer-drug relationship, thereby improving the accuracy of probability output and better predicting cancer-drug relationships. Therefore, the method for analyzing the relationship between drugs and cancer based on graph neural networks proposed in this invention can improve the analysis effect of the relationship between drugs and cancer by updating historical data vectors to obtain new cancer vectors and drug vectors, and by using the graph structure of cancer information and drug information as input data for the graph neural network model.

[0092] like Figure 2 The diagram shown is a functional block diagram of the device for analyzing the relationship between drugs and cancer based on graph neural networks according to the present invention.

[0093] The graph neural network-based drug-cancer relationship analysis device 100 of this invention can be installed in an electronic device. Depending on the functions implemented, the graph neural network-based drug-cancer relationship analysis device may include a vector determination module 101, a first loss calculation module 102, a second loss calculation module 103, and a relationship recognition module 104. The module described in this invention can also be called a unit, which refers to a series of computer program segments that can be executed by the processor of an electronic device and can perform a fixed function, and are stored in the memory of the electronic device.

[0094] In this embodiment, the functions of each module / unit are as follows:

[0095] The vector determination module 101 is used to acquire cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, the module uses a graph neural network model to determine the cancer vector of the cancer information and the graph neural network model to determine the drug vector of the drug information.

[0096] The first loss calculation module 102 is used to construct a cancer-drug bipartite graph between the cancer information and the drug information, and to calculate the first loss value of the graph neural network model using the cancer vector, the drug vector and the cancer-drug bipartite graph;

[0097] The second loss calculation module 103 is used to calculate a second loss value between the cancer information and the drug information based on the cancer vector, the drug vector and the cancer-drug bipartite graph;

[0098] The relationship recognition module 104 is used to train the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model, and to use the trained graph neural network model to identify the cancer-drug relationship between the cancer information and the drug information.

[0099] In detail, the modules in the graph neural network-based drug-cancer relationship analysis device 100 described in this embodiment of the invention employ the same methods as described above during use. Figure 1 The method described herein is the same as the graph neural network-based drug-cancer relationship analysis method and can produce the same technical effect, so it will not be elaborated here.

[0100] like Figure 3 The diagram shown is a schematic diagram of the electronic device 1 that implements the method for analyzing the relationship between drugs and cancer based on graph neural networks according to the present invention.

[0101] The electronic device 1 may include a processor 10, a memory 11, a communication bus 12, and a communication interface 13. It may also include a computer program stored in the memory 11 and capable of running on the processor 10, such as a program for analyzing the relationship between drugs and cancer based on graph neural networks.

[0102] In some embodiments, the processor 10 may be composed of integrated circuits, such as a single packaged integrated circuit or multiple integrated circuits with the same or different functions, including combinations of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips. The processor 10 is the control unit of the electronic device 1, connecting various components of the electronic device 1 via various interfaces and lines. It executes programs or modules stored in the memory 11 (e.g., executing a program for analyzing the relationship between drugs and cancer based on graph neural networks) and calls data stored in the memory 11 to perform various functions and process data of the electronic device 1.

[0103] The memory 11 includes at least one type of readable storage medium, including flash memory, portable hard drive, multimedia card, card-type memory (e.g., SD or DX memory), magnetic memory, disk, optical disk, etc. In some embodiments, the memory 11 can be an internal storage unit of the electronic device 1, such as a portable hard drive of the electronic device 1. In other embodiments, the memory 11 can be an external storage device of the electronic device 1, such as a plug-in portable hard drive, smart media card (SMC), secure digital card (SD), flash card, etc., equipped on the electronic device 1. Furthermore, the memory 11 can include both internal and external storage units of the electronic device 1. The memory 11 can be used not only to store application software and various types of data installed on the electronic device 1, such as code for a drug-cancer relationship analysis program based on graph neural networks, but also to temporarily store data that has been output or will be output.

[0104] The communication bus 12 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. This bus can be divided into an address bus, a data bus, a control bus, etc. The bus is configured to enable communication between the memory 11 and at least one processor 10, etc.

[0105] The communication interface 13 is used for communication between the aforementioned electronic device 1 and other devices, including a network interface and an employee interface. Optionally, the network interface may include a wired interface and / or a wireless interface (such as a Wi-Fi interface, Bluetooth interface, etc.), typically used to establish a communication connection between the electronic device 1 and other electronic devices 1. The employee interface may be a display, an input unit (such as a keyboard), or, optionally, a standard wired or wireless interface. Optionally, in some embodiments, the display may be an LED display, a liquid crystal display, a touch-sensitive liquid crystal display, or an OLED (Organic Light-Emitting Diode) touchscreen, etc. The display may also be appropriately referred to as a screen or display unit, used to display information processed in the electronic device 1 and to display a visual employee interface.

[0106] Figure 3 Only electronic device 1 with components is shown; those skilled in the art will understand that... Figure 3 The structure shown does not constitute a limitation on the electronic device 1, and may include fewer or more components than shown, or combine certain components, or have different component arrangements.

[0107] For example, although not shown, the electronic device 1 may also include a power supply (such as a battery) to power various components. Preferably, the power supply can be logically connected to the at least one processor 10 through a power management device, thereby enabling functions such as charging management, discharging management, and power consumption management. The power supply may also include one or more DC or AC power supplies, recharging devices, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components. The electronic device 1 may also include various sensors, Bluetooth modules, Wi-Fi modules, etc., which will not be described in detail here.

[0108] It should be understood that the embodiments described are for illustrative purposes only and are not limited to this structure in terms of the scope of the patent invention.

[0109] The graph neural network-based drug-cancer relationship analysis program stored in the memory 11 of the electronic device 1 is a combination of multiple computer programs. When run in the processor 10, it can achieve the following:

[0110] Obtain cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, use a graph neural network model to determine the cancer vector of the cancer information, and use the graph neural network model to determine the drug vector of the drug information.

[0111] Construct a cancer-drug bipartite graph between the cancer information and the drug information, and use the cancer vector, the drug vector, and the cancer-drug bipartite graph to calculate the first loss value of the graph neural network model;

[0112] Based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, calculate a second loss value between the cancer information and the drug information;

[0113] The graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model. The trained graph neural network model is then used to identify the cancer-drug relationship between the cancer information and the drug information.

[0114] Specifically, the processor 10's implementation method of the above-mentioned computer program can be found in [reference needed]. Figure 1 The descriptions of the relevant steps in the corresponding embodiments are not repeated here.

[0115] Furthermore, if the modules / units integrated in the electronic device 1 are implemented as software functional units and sold or used as independent products, they can be stored in a non-volatile computer-readable storage medium. The computer-readable storage medium can be volatile or non-volatile. For example, the computer-readable medium may include: any entity or device capable of carrying the computer program code, a recording medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, or a read-only memory (ROM).

[0116] The present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor of an electronic device 1, can perform the following:

[0117] Obtain cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, use a graph neural network model to determine the cancer vector of the cancer information, and use the graph neural network model to determine the drug vector of the drug information.

[0118] Construct a cancer-drug bipartite graph between the cancer information and the drug information, and use the cancer vector, the drug vector, and the cancer-drug bipartite graph to calculate the first loss value of the graph neural network model;

[0119] Based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, calculate a second loss value between the cancer information and the drug information;

[0120] The graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model. The trained graph neural network model is then used to identify the cancer-drug relationship between the cancer information and the drug information.

[0121] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of modules is only a logical functional division, and other division methods may be used in actual implementation.

[0122] The modules described as separate components may or may not be physically separate. The components shown as modules may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs.

[0123] Furthermore, the functional modules in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or in the form of hardware plus software functional modules.

[0124] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0125] Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description. Thus, all variations falling within the meaning and scope of equivalents of the claims are intended to be embraced within the invention. No appended diagram markings in the claims should be construed as limiting the scope of the claims.

[0126] The embodiments of this invention can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that utilize digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0127] Furthermore, it is clear that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices recited in a system claim may also be implemented by a single unit or device through software or hardware. The term "second class" is used to indicate names and does not indicate any specific order.

[0128] Finally, it should be noted that 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 preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for analyzing the relationship between drugs and cancer based on graph neural networks, characterized in that, The method includes: Obtain cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, use a graph neural network model to determine the cancer vector of the cancer information, and use the graph neural network model to determine the drug vector of the drug information. Construct a cancer-drug bipartite graph between the cancer information and the drug information, and use the cancer vector, the drug vector, and the cancer-drug bipartite graph to calculate the first loss value of the graph neural network model; Based on the cancer vector, the drug vector, and the cancer-drug bipartite graph, calculate a second loss value between the cancer information and the drug information; The graph neural network model is trained using the first loss value and the second loss value to obtain a trained graph neural network model. The trained graph neural network model is then used to identify the cancer-drug relationship between the cancer information and the drug information. The step of calculating a second loss value between cancer information and drug information based on the cancer vector, the drug vector, and the cancer-drug bipartite graph includes: performing bipartite graph encoding on the cancer-drug bipartite graph to obtain a bipartite graph vector; constructing initial node vectors for the cancer-drug bipartite graph using the cancer vector and the drug vector; updating the initial node vectors to obtain updated node vectors; scrambling the structure of the cancer-drug bipartite graph to obtain a scrambled bipartite graph; calculating the scrambled node vectors of the scrambled bipartite graph; calculating a first vector distance between the scrambled node vectors and the bipartite graph vectors; calculating a second vector distance between the updated node vectors and the bipartite graph vectors; and calculating the second loss value using the following formula based on the first vector distance and the second vector distance: in, This represents the second loss value. Represents the distance of the first vector, i.e., the distance of the first vector. The shuffled node vector and the bipartite graph vector in the th _ ... The distance between the vectors of the nodes. Represents the distance of the second vector, i.e., the first vector. The updated node vector and the bipartite graph vector in the i-th update node vector and the i-th node vector in the bipartite graph. The distance between the vectors of the nodes. This represents the number of nodes in a bipartite graph. This represents the node number in the bipartite graph.

2. The method for analyzing the relationship between drugs and cancer based on graph neural networks as described in claim 1, characterized in that, The multi-omics data and gene regulatory network used to query the cancer information include: Identify the target of the cancer information query; Based on the query objective, query the multi-omics data of the cancer information; Based on the multi-omics data, a prior distribution of the regulatory network for the cancer information is constructed; A gene regulation network model for cancer information is constructed using the prior distribution of the regulatory network, and the gene regulation network model is used as the gene regulation network.

3. The method for analyzing the relationship between drugs and cancer based on graph neural networks as described in claim 1, characterized in that, The molecular map for querying the drug information includes: Determine the name of the drug in the drug information; To find the drug components of the drug name; Extract the molecular topological structure of the drug's constituent molecules; The molecular topology is used to determine the molecular map of the drug information.

4. The method for analyzing the relationship between drugs and cancer based on graph neural networks as described in claim 1, characterized in that, The construction of the cancer-drug bipartite graph between the cancer information and the drug information includes: Determine the cancer vertex of the cancer information and the drug vertex of the drug information; Query the historical relationship between the cancer vertex and the drug vertex; Based on the historical relationships, construct the relationship edges between the cancer vertex and the drug vertex; Based on the relation edges, the cancer vertex, and the drug vertex, a cancer-drug bipartite graph is determined between the cancer information and the drug information.

5. The method for analyzing the relationship between drugs and cancer based on graph neural networks as described in claim 1, characterized in that, The step of calculating the first loss value of the graph neural network model using the cancer vector, the drug vector, and the cancer-drug bipartite graph includes: Calculate the dot product between the cancer vector and the drug vector; Calculate the activation probability value of the vector inner product; In the cancer-drug bipartite graph, query the true relationship between the cancer vector and the drug vector; Determine the probability of the true relationship. Calculate the cross-entropy loss value between the true relation probability and the activation probability value, and use the cross-entropy loss value as the first loss value of the graph neural network model.

6. The method for analyzing the relationship between drugs and cancer based on graph neural networks as described in claim 1, characterized in that, The step of training the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model includes: The graph neural network model is backpropagated using the first loss value and the second loss value to obtain the backpropagation model. Calculate the output loss value of the backpropagation model; When the output loss value is not less than the preset loss value, return to the above steps of using the first loss value and the second loss value to perform backpropagation processing on the graph neural network model to obtain the backpropagation model; When the output loss value is less than the preset loss value, the trained graph neural network model is obtained.

7. A device for analyzing the relationship between drugs and cancer based on graph neural networks, used to implement the method for analyzing the relationship between drugs and cancer based on graph neural networks as described in any one of claims 1 to 6, characterized in that, The device includes: The vector determination module is used to acquire cancer information and drug information, query the multi-omics data and gene regulatory network of the cancer information, and query the molecular map of the drug information. Based on the multi-omics data and the gene regulatory network, the module uses a graph neural network model to determine the cancer vector of the cancer information and the graph neural network model to determine the drug vector of the drug information. The first loss calculation module is used to construct a cancer-drug bipartite graph between the cancer information and the drug information, and to calculate the first loss value of the graph neural network model using the cancer vector, the drug vector and the cancer-drug bipartite graph. The second loss calculation module is used to calculate a second loss value between the cancer information and the drug information based on the cancer vector, the drug vector and the cancer-drug bipartite graph; The relationship recognition module is used to train the graph neural network model using the first loss value and the second loss value to obtain a trained graph neural network model, and to use the trained graph neural network model to identify the cancer-drug relationship between the cancer information and the drug information.

8. An electronic device, characterized in that, The electronic device includes: At least one processor; and, A memory communicatively connected to the at least one processor; wherein, The memory stores a computer program that can be executed by the at least one processor, the computer program being executed by the at least one processor to enable the at least one processor to perform the method for analyzing the relationship between drugs and cancer based on graph neural networks as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for analyzing the relationship between drugs and cancer based on graph neural networks as described in any one of claims 1 to 6.

Images