Drug-target interaction prediction method and device based on symbolic graph neural network

Abstract

The present invention provides a drug-target interaction prediction method and device based on a symbolic graph neural network, the method comprising: obtaining a drug-target symbolic heterogeneous graph, and obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; using a symbolic graph neural network model to process the symbolic bipartite network to obtain embedding results of the drug and target; performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; splicing the embedding results with the auxiliary information to obtain the final embedding results of the drug and target, and applying a discriminator to predict the symbols between the drug and the target to obtain the drug-target interaction prediction results. In the above manner, the present invention can achieve a higher prediction accuracy by classifying the specific patterns of interaction between the drug and the target and constructing a symbolic heterogeneous graph, and then applying the symbolic graph neural network model for processing.

Description

Technical Field

[0001] The present invention belongs to the field of medical technology, and specifically relates to a drug-target interaction prediction method and device based on a symbolic graph neural network. Background Art

[0002] Predicting drug-target interactions (DTIs) is crucial for both drug discovery and drug repositioning. However, the rapid increase in the number of drugs and targets makes traditional biological experimental methods time-consuming and expensive. Computational methods have gained increasing attention for accurate DTI prediction and could significantly reduce the time and cost of drug discovery and drug repositioning.

[0003] Numerous approaches are currently being developed for predicting drug-target relationships. When addressing the problem of predicting drug-target relationships, this problem is often transformed into a link prediction problem. However, most existing DTI prediction methods ignore the specific interaction modes between drugs and targets. By querying the biological database DrugBank, we can discover that drug-target interactions involve numerous mechanisms. These interactions can be broadly categorized as positive and negative relationships, and are naturally represented as symbolic links.

[0004] In recent years, graph neural networks (GNNs) have demonstrated excellent performance in a wide range of graph mining applications, including drug discovery. GNN-based methods naturally combine multiple data sources into a single DTI prediction framework. However, most GNN-based models fail to leverage the symbolically heterogeneous nature of DTI network data. Summary of the Invention

[0005] The present invention provides a drug-target interaction prediction method and device based on a symbolic graph neural network, which solves the problem that existing drug-target interaction prediction ignores the drug-target interaction relationship. By classifying the specific patterns of interaction between drugs and targets and constructing a symbolic heterogeneous graph, and then applying a symbolic graph neural network model for processing, a higher prediction accuracy can be achieved.

[0006] Based on the above objectives, the present invention proposes a drug-target interaction prediction method based on a symbolic graph neural network, comprising: obtaining a drug-target symbolic heterogeneous graph, and obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; using a symbolic graph neural network model to process the symbolic bipartite network to obtain embedding results of drugs and targets; performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; splicing the embedding results with the auxiliary information to obtain the final embedding results of the drug and target, and applying a discriminator to predict the symbols between the drug and the target to obtain a drug-target interaction prediction result.

[0007] Optionally, the obtaining of the drug-target symbol heterogeneous graph includes: collecting drug-target interaction relationships, drug characteristics, drug interaction relationships and target interaction relationships; incorporating the modes of action of drugs and targets into symbol classification, and constructing a drug-target symbol heterogeneous graph in combination with the collected interaction relationships.

[0008] Optionally, obtaining a symbolic bipartite network, a drug interaction network and a target interaction network based on the drug-target symbolic heterogeneous graph includes: extracting drugs, targets and drug-target interaction relationships from the drug-target symbolic heterogeneous graph to obtain a symbolic bipartite network; extracting drugs and drug interaction relationships from the drug-target symbolic heterogeneous graph to obtain a drug interaction network; extracting targets and target interaction relationships from the drug-target symbolic heterogeneous graph to obtain a target interaction network.

[0009] Optionally, the use of a symbolic graph neural network model to process a symbolic bipartite network to obtain embedding results of drugs and targets includes: applying a graph convolution layer based on symbolic relationships in the symbolic graph neural network model to iteratively calculate the embedding process of drugs and targets, obtaining the positive and negative neighborhood embedding results of the drug and the positive and negative neighborhood embedding results of the target in the symbolic bipartite network; splicing the positive and negative neighborhood embedding results of the drug and the positive and negative neighborhood embedding results of the target to obtain the embedding results of the drug and the target.

[0010] Optionally, performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information includes: performing principal component analysis on the drug interaction network to obtain first auxiliary information corresponding to the drug interaction network; performing principal component analysis on the target interaction network to obtain second auxiliary information corresponding to the target interaction network.

[0011] Optionally, the embedding result includes a drug embedding result and a target embedding result, and the embedding result is spliced with the auxiliary information to obtain the final embedding result of the drug and the target, including: splicing the drug embedding result with the first auxiliary information to obtain the final drug embedding result; splicing the target embedding result with the second auxiliary information to obtain the final target embedding result.

[0012] Optionally, the final embedding result includes a final drug embedding result and a final target embedding result, the discriminator is a softmax regression classifier, and the application of the discriminator to predict the symbols between the drug and the target to obtain the drug-target interaction prediction result includes: the final drug embedding result and the final target embedding result are input into the softmax regression classifier in pairs; the softmax regression classifier is applied to predict the symbols between the drug and the target to obtain the drug-target interaction prediction result.

[0013] Based on the same inventive concept, the present invention also proposes a drug-target interaction prediction device based on a symbolic graph neural network, comprising: a network generation unit, used to obtain a drug-target symbolic heterogeneous graph, and obtain a symbolic bipartite network, a drug interaction network and a target interaction network according to the drug-target symbolic heterogeneous graph; an embedding result acquisition unit, used to process the symbolic bipartite network using a symbolic graph neural network model to obtain embedding results of drugs and targets; a component analysis unit, used to perform principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; a result prediction unit, used to splice the node embedding result with the auxiliary information, and apply a discriminator to predict the symbols between the drug and the target to obtain a drug-target interaction prediction result.

[0014] Based on the same inventive concept, the present invention also proposes an electronic device, including a memory, a processor, and a computer program stored in the memory and runnable on the processor. When the processor executes the program, the drug-target interaction prediction method based on the symbolic graph neural network as described above is implemented.

[0015] Based on the same inventive concept, the present invention also proposes a computer storage medium, which stores at least one executable instruction, and the executable instruction enables the processor to execute the drug-target interaction prediction method based on the symbolic graph neural network as described above.

[0016] From the above description, it can be seen that the beneficial effects of the technical solution provided by the present invention are: the present invention provides a drug-target interaction prediction method and device based on a symbolic graph neural network, the method comprising: obtaining a drug-target symbolic heterogeneous graph, and obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; obtaining the embedding results of the drug and target based on the symbolic bipartite network after application training; performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; splicing the embedding results with the auxiliary information to obtain the final embedding results of the drug and target, and applying a discriminator to predict the symbols between the drug and the target to obtain the drug-target interaction prediction results, and by classifying the specific patterns of the interaction between the drug and the target and constructing a symbolic heterogeneous graph, and then applying the symbolic graph neural network model for processing, a higher prediction accuracy can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0018] Figure 1 Schematic diagram of the process of drug-target interaction prediction method based on signed graph neural network according to an embodiment of the present invention;

[0019] Figure 2 Schematic diagram of the structure of a signed graph neural network according to an embodiment of the present invention;

[0020] Figure 3 Schematic diagram of the structure of a drug-target interaction prediction device based on a signed graph neural network according to an embodiment of the present invention;

[0021] Figure 4 FIG. 4 is a schematic diagram of the hardware structure of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0022] In order to make the objectives, technical solutions and advantages of the present disclosure more clearly understood, the present disclosure is further described in detail below in conjunction with specific embodiments and with reference to the accompanying drawings.

[0023] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the embodiments of the present invention should have the usual meanings understood by people with ordinary skills in the field to which the present disclosure belongs. The "first", "second" and similar words used in the embodiments of the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0024] The present invention implements a drug-target interaction prediction method based on a signed graph neural network. Figure 1 As shown in Figure 2, the drug-target interaction prediction method based on symbolic graph neural network includes:

[0025] Step S11: obtaining a drug-target symbolic heterogeneous graph, and obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph.

[0026] Optionally, drug-target interaction relationships, drug characteristics, drug interaction relationships, and target interaction relationships are collected; the modes of action of the drug and target are incorporated into the symbol classification, and a drug-target symbolic isomorphic graph is constructed based on the collected interaction relationships. Drug characteristics refer to the chemical structure information of the drug. Targets are biological macromolecules in the body that have pharmacodynamic functions and can be acted upon by drugs, such as certain proteins and nucleic acids. In this application, the targets are human proteins, i.e., proteins.

[0027] In an embodiment of the present invention, known human protein interaction data are searched in the database String to obtain interaction information between targets.

[0028] The chemical structure information of the drug was obtained from the biological database PubChem. is the chemical structure feature matrix of the drug node, n is the number of drugs, d D is the dimension of the drug, preferably 881, that is, each drug is represented by an 881-dimensional binary vector, where a value of 1 indicates that the drug has a specific chemical structure segment, and a value of 0 indicates that the drug does not have the chemical structure segment.

[0029] The modes of action of drugs and targets are represented as symbolic links, where types such as "agonist" and "activator" are classified as positive and labeled "+1," while types such as "inhibitor" and "antagonist" are classified as negative and labeled "-1." Some categories, such as "modulators," "binder," and "cleavage," cannot be included in the symbolic classification. In these cases, no edge is constructed between the drug and target. The result is a two-layer drug-target symbolic heterogeneous network.

[0030] When obtaining the symbolic bipartite network, the drug interaction network and the target interaction network according to the drug-target symbolic heterogeneous graph, it is optional, such as Figure 2 As shown, drugs, targets, and drug-target interaction relationships are extracted from the drug-target symbolic heterogeneous graph to obtain a symbolic bipartite network; drugs and drug interaction relationships are extracted from the drug-target symbolic heterogeneous graph to obtain a drug interaction network; targets and target interaction relationships are extracted from the drug-target symbolic heterogeneous graph to obtain a target interaction network. Specifically, in the drug-target symbolic heterogeneous graph, the drug node set D = (d1, d2, ..., d n ), target node set T={t1,t2,…,t m}, the two interaction relationships between drugs and targets are processed as symbolic edge sets E DT ={e ij , i=1…n, j=1…m}. The drug-target relationship is expressed as a symbolic bipartite graph G=(D,T,E DT ). Drug interaction information and target interaction information are used to construct drug interaction networks (DDIs) and target interaction networks (PPIs). The expression of drug interaction network (DDIs) is G D =(D,A D ,E D ), the expression of target interaction network (PPIs) is G T =(D,A T ,E T ), where the adjacency matrix A D Indicates whether there is interaction between drug nodes, adjacency matrix A T Indicates whether there is interaction between target node pairs.

[0031] Step S12: Use the symbolic graph neural network model to process the symbolic bipartite network to obtain the embedding results of drugs and targets.

[0032] In an embodiment of the present invention, a symbolic relation-based GNN model, referred to as an SRGNN, is used to study node embedding results on a drug-target symbolic heterogeneous graph. Optionally, a symbolic relation-based graph convolutional layer in a symbolic graph neural network model is used to iteratively calculate the embedding process of the drug and target, obtaining the positive and negative neighborhood embedding results of the drug and the target in the symbolic bipartite network; the positive and negative neighborhood embedding results of the drug and the target are concatenated to obtain the embedding results of the drug and target.

[0033] make and is the input of the lth layer of SRGNN, and are the positive neighbors of the drug and negative neighborhood The hidden representation of the lth layer, that is, the positive and negative neighborhood embedding results of the drug. and are the positive neighbors of the target and negative neighborhood The hidden representation of the lth layer, that is, the positive and negative neighborhood embedding results of the target, σ is a nonlinear activation function, is the learnable network weight parameter matrix. The process of iteratively calculating the drug node embedding using SRGNN is as follows:

[0034]

[0035]

[0036] Similarly, the target node also has a similar embedding process:

[0037]

[0038]

[0039] Among them, l∈{1,2,…,L}, i∈{1,2,…,n}, j∈{1,2,…,m}, and L is the number of convolutional layers.

[0040] The positive and negative neighborhood embedding results of the drug and the target are concatenated to obtain the embedding results of the drug and target as the input of the next layer:

[0041]

[0042]

[0043] in, is the drug embedding result of the lth layer, is the target embedding result of the lth layer.

[0044]

[0045]

[0046] and It is the embedding result of drugs and targets obtained by L-layer SRGNN on the symbolic bipartite graph.

[0047] Step S13: performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information.

[0048] Optionally, principal component analysis is performed on the drug interaction network to obtain first auxiliary information corresponding to the drug interaction network: Z DDI ←PCA(A D ), Perform principal component analysis on the target interaction network to obtain the second auxiliary information corresponding to the target interaction network: Z PPI ←PCA(A T ). Here, PCA stands for principal component analysis.

[0049] Step S14: The embedding result is concatenated with the auxiliary information to obtain the final embedding result of the drug and the target, and a discriminator is applied to predict the symbols between the drug and the target to obtain the drug-target interaction prediction result.

[0050] The final embedding results include the final drug embedding results Z D And the final target embedding result Z T In step S14, the drug embedding result is spliced with the first auxiliary information to obtain the final drug embedding result. The target embedding result is spliced with the second auxiliary information to obtain the final target embedding result. A||B=(A,B), then the final drug embedding result Final target embedding results Among them, s is the dimension of embedding and auxiliary information before splicing operation.

[0051] In an embodiment of the present invention, the discriminator is preferably a softmax regression classifier. In step S14, the final drug embedding result and the final target embedding result are input into the softmax regression classifier in pairs; the softmax regression classifier is applied to predict the sign between the drug and the target to obtain the drug-target interaction prediction result. Specifically, for the final drug embedding result Z D And the final target embedding result ZT , the drug d i With target t j The embedding result of The data is input into the softmax regression classifier to predict the symbols between the drug and the target, and the drug-target interaction prediction results are obtained.

[0052] In the embodiment of the present invention, the SRGNN and the discriminator are jointly trained using the loss function described below. The loss function used to jointly train the model and the discriminator, the softmax regression classifier has a regression coefficient in, is the positive edge type coefficient, is the negative edge type coefficient. Let Θ NN ={Θ NN(1) ,…,Θ NN(L)} is the weight matrix parameter of SRGNN, and the loss function is defined as follows:

[0053]

[0054] Among them, e ij ∈S represents the type of edge between drug and target, S represents the set of edges between drug and target, Indicates the edge type e ij The associated weights, R is the regression coefficient space of the softmax regression classifier, and Π(·) returns 1 if the given prediction is true, otherwise it returns 0.

[0055] The drug-target interaction prediction method based on symbolic graph neural network in an embodiment of the present invention classifies the specific patterns of interaction between drugs and targets, and additionally extracts the interactions between drug pairs and target pairs, models the drug-target interaction DTIs as a two-layer symbolic heterogeneous network, proposes a symbolic graph neural network model based on relational graph convolution, constructs a two-layer data graph model of SRGNN and discriminator, and realizes DTIs symbolic prediction.

[0056] The following tests the predictive performance of the drug-target interaction prediction method based on a symbolic graph neural network (SGN) according to an embodiment of the present invention, examining the number of convolutional layers L = {1, 2, 3, 4, 5} and the embedding dimension s = {4, 8, 16, 32, 128, 256}. The SRGNN uses an Adam optimizer with a learning rate of 0.01. The model is trained using different numbers of iterations, and it was found that 200 iterations achieves excellent results. The SGN-based drug-target interaction prediction method according to an embodiment of the present invention also considers several initial features of drug and target nodes, as well as whether to integrate auxiliary information extracted by principal component analysis. The recently developed logistic regression method on symbolic heterogeneous networks (SCsc) and random walk method (SBRW) are used as comparative examples. SCsc is a feature extraction method guided by equilibrium theory that extracts edge features by calculating the number of equilibrium paths. The area under the receiver operating characteristic curve (AUC) and the F1 metric are used as evaluation metrics, with higher values indicating better model performance. Each experimental run uses 5-fold cross validation, and all results are the average of 5 runs. The prediction results of different methods are shown in Table 1.

[0057] Table 1 Statistics of optimal indicators of different methods on the extracted data sets

[0058] method AUC F1 SCsc 0.854 0.805 SBRW 0.824 0.757 SRGNN 0.895 0.860

[0059] Table 1 shows the optimal indicator statistics of different methods. It is obvious that the performance of the drug-target interaction prediction method based on the symbolic graph neural network in the embodiment of the present invention greatly exceeds the logistic regression method (SCsc) and the random walk method (SBRW).

[0060] In summary, the drug-target interaction prediction method based on a symbolic graph neural network in an embodiment of the present invention obtains a drug-target symbolic heterogeneous graph, and obtains a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; obtains the embedding results of the drug and target based on the symbolic bipartite network after training; performs principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; splices the embedding results with the auxiliary information to obtain the final embedding results of the drug and target, and applies a discriminator to predict the symbols between the drug and the target to obtain the drug-target interaction prediction results, and constructs a symbolic heterogeneous graph by classifying the specific patterns of interaction between the drug and the target and additionally obtaining information on drug pairs and target pairs, and then uses the symbolic graph neural network model and the principal component analysis method to process the symbolic heterogeneous graph, which can achieve a higher prediction accuracy.

[0061] The foregoing description is of specific embodiments of the present invention. In some cases, the actions or steps described in the embodiments of the present invention may be performed in an order different from that shown in the embodiments and still achieve the desired results. In addition, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0062] The embodiment of the present invention also provides a drug-target interaction prediction device based on a signed graph neural network, such as Figure 3 As shown, the drug-target interaction prediction device based on the symbolic graph neural network includes: a network generation unit, an embedding result acquisition unit, a component analysis unit and a result prediction unit.

[0063] A network generation unit, configured to obtain a drug-target symbolic heterogeneous graph, and to obtain a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph;

[0064] An embedding result acquisition unit, used for processing the symbolic bipartite network using a symbolic graph neural network model to obtain embedding results of drugs and targets;

[0065] a component analysis unit, configured to perform principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information;

[0066] The result prediction unit is used to splice the node embedding result with the auxiliary information, and use a discriminator to predict the sign between the drug and the target to obtain a drug-target interaction prediction result.

[0067] For the convenience of description, the above device is described as being divided into various units according to their functions. Of course, when implementing the embodiments of the present invention, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0068] The apparatus of the above embodiment is applied to the corresponding method of the above embodiment and has the beneficial effects of the corresponding method embodiment, which will not be described in detail here.

[0069] Based on the same inventive concept, an embodiment of the present invention further provides an electronic device, which includes a memory, a processor, and a computer program stored in the memory and runnable on the processor, wherein when the processor executes the program, the method described in any one of the above embodiments is implemented.

[0070] An embodiment of the present invention provides a non-volatile computer storage medium, wherein the computer storage medium stores at least one executable instruction, and the computer executable instruction can execute the method described in any one of the above embodiments.

[0071] Figure 4 A more specific hardware structure diagram of an electronic device provided in this embodiment is shown. The device may include: a processor 401, a memory 402, an input / output interface 403, a communication interface 404, and a bus 405. The processor 401, the memory 402, the input / output interface 403, and the communication interface 404 are communicatively connected to each other within the device via the bus 405.

[0072] The processor 401 can be implemented as a general-purpose CPU (Central Processing Unit), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided by the method embodiments of the present invention.

[0073] The memory 402 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage devices, dynamic storage devices, etc. The memory 402 can store an operating system and other application programs. When the technical solutions provided by the method embodiments of the present invention are implemented through software or firmware, the relevant program codes are stored in the memory 402 and called and executed by the processor 401.

[0074] The input / output interface 403 is used to connect to input / output modules to implement information input and output. The input / output modules can be configured as components within the device (not shown in the figure) or can be externally connected to the device to provide corresponding functions. Input devices may include a keyboard, mouse, touch screen, microphone, various sensors, etc., and output devices may include a display, speaker, vibrator, indicator light, etc.

[0075] The communication interface 404 is used to connect to a communication module (not shown) to enable communication between the device and other devices. The communication module can communicate via a wired method (such as USB, network cable, etc.) or a wireless method (such as mobile network, WIFI, Bluetooth, etc.).

[0076] The bus 405 comprises a pathway for transmitting information between various components of the device (eg, the processor 401 , the memory 402 , the input / output interface 403 , and the communication interface 404 ).

[0077] It should be noted that although the above device only shows the processor 401, memory 402, input / output interface 403, communication interface 404, and bus 405, in a specific implementation, the device may also include other components necessary for normal operation. In addition, those skilled in the art will understand that the above device may only include the components necessary to implement the embodiments of the present invention, and does not necessarily include all the components shown in the figure.

[0078] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present disclosure (including the claims) is limited to these examples. Within the scope of the present disclosure, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the embodiments of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0079] The embodiments of the present invention are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the scope of protection of this disclosure.

Claims

1. A drug-target interaction prediction method based on a symbolic graph neural network, characterized by: The drug-target interaction prediction method comprises: Obtaining a drug-target symbolic heterogeneous graph, and obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; Processing the symbolic bipartite network using a symbolic graph neural network model to obtain embedding results of the drug and the target, including: applying a graph convolution layer based on symbolic relationships in the symbolic graph neural network model to iteratively calculate the embedding process of the drug and the target to obtain positive and negative neighborhood embedding results of the drug and positive and negative neighborhood embedding results of the target in the symbolic bipartite network; splicing the positive and negative neighborhood embedding results of the drug and the positive and negative neighborhood embedding results of the target to obtain embedding results of the drug and the target; performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; The embedding result is concatenated with the auxiliary information to obtain the final embedding result of the drug and the target, and a discriminator is applied to predict the sign between the drug and the target to obtain the drug-target interaction prediction result.

2. The drug-target interaction prediction method according to claim 1, wherein: The obtaining of a drug-target symbol isomerism graph includes: Collect drug-target interactions, drug characteristics, drug interactions, and target interactions; The modes of action of drugs and targets are incorporated into the symbol classification, and the collected interaction relationships are combined to construct a drug-target symbol heterogeneous graph.

3. The drug-target interaction prediction method according to claim 1, wherein: The step of obtaining a symbolic bipartite network, a drug interaction network, and a target interaction network according to the drug-target symbolic heterogeneous graph includes: Extracting drugs, targets, and drug-target interaction relationships from the drug-target symbolic heterogeneous graph to obtain a symbolic bipartite network; extracting drugs and drug interaction relationships from the drug-target symbolic heterogeneous graph to obtain a drug interaction network; Targets and target interaction relationships are extracted from the drug-target symbolic heterogeneous graph to obtain a target interaction network.

4. The drug-target interaction prediction method according to claim 1, wherein: The performing principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information includes: performing principal component analysis on the drug interaction network to obtain first auxiliary information corresponding to the drug interaction network; A principal component analysis is performed on the target interaction network to obtain second auxiliary information corresponding to the target interaction network.

5. The drug-target interaction prediction method according to claim 4, wherein: The embedding result includes a drug embedding result and a target embedding result. The embedding result and the auxiliary information are spliced to obtain a final embedding result of the drug and the target, including: splicing the drug embedding result with the first auxiliary information to obtain a final drug embedding result; The target embedding result is concatenated with the second auxiliary information to obtain a final target embedding result.

6. The drug-target interaction prediction method according to claim 1, wherein: The final embedding result includes a final drug embedding result and a final target embedding result. The discriminator is a softmax regression classifier. The application of the discriminator to predict the sign between the drug and the target to obtain the drug-target interaction prediction result includes: The final drug embedding result and the final target embedding result are input into the softmax regression classifier in pairs; The softmax regression classifier is applied to predict the sign between the drug and the target to obtain the drug-target interaction prediction result.

7. A drug-target interaction prediction device based on a signed graph neural network, characterized in that: The drug-target interaction prediction device comprises: A network generation unit, configured to obtain a drug-target symbolic heterogeneous graph, and to obtain a symbolic bipartite network, a drug interaction network, and a target interaction network based on the drug-target symbolic heterogeneous graph; An embedding result acquisition unit uses a symbolic graph neural network model to process a symbolic bipartite network to obtain embedding results of the drug and the target, including: applying a graph convolution layer based on symbolic relations in the symbolic graph neural network model to iteratively calculate the embedding process of the drug and the target, and obtaining the positive and negative neighborhood embedding results of the drug and the target in the symbolic bipartite network; splicing the positive and negative neighborhood embedding results of the drug and the target to obtain the embedding results of the drug and the target; a component analysis unit, configured to perform principal component analysis on the drug interaction network and the target interaction network to obtain auxiliary information; A result prediction unit is used to splice the embedding result with the auxiliary information, and use a discriminator to predict the sign between the drug and the target to obtain a drug-target interaction prediction result.

8. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, the method according to any one of claims 1 to 6 is implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images