Drug relocation method based on multistage cross-domain relation network

Abstract

The invention discloses a drug relocation method based on a multi-level cross-domain relation network, and belongs to the technical field of drug relocation, the multi-level cross-domain relation network including a drug-disease associated native network, a drug associated network and a disease associated network is constructed, and multiple network architectures are fused to enable the network to reveal richer semantic relations, so that the drug-disease associated native network and the drug-disease associated network are integrated, and the drug-disease associated native network and the drug-disease associated network are integrated; the expression capability of the network is enhanced; a cross-domain fusion encoder is designed, a multi-level cross-domain relation network is encoded for a drug relocation task, and the accuracy of network encoding is remarkably improved by comprehensively encoding multi-source heterogeneous data; a multilayer sensor is used as a decoder to predict association between drugs and diseases; and forming a drug relocation model by the multi-level cross-domain relation network, the cross-domain fusion encoder and the multi-layer sensor, constructing a classification task to train the model, inputting drugs and diseases for testing into the trained drug relocation model, and outputting a probability score of the curative effect of the tested drugs on the tested diseases.

Description

Technical Field

[0001] The present invention belongs to the technical field of drug repositioning, and in particular relates to a drug repositioning method based on a multi-level cross-domain relationship network. Background Art

[0002] In the process of drug discovery, traditional drug development often requires decades of clinical research and clinical trials, and then it can be marketed after market approval, which consumes huge manpower and financial resources. However, studies have shown that marketed drugs exhibit therapeutic effects different from the original indications in cytology, animal models or patient treatment. For example, Remdesivir, used to treat Ebola virus, has been proven to provide a rapid response in the treatment of COVID-19, or Fluoxetine was originally used to treat depression and was later found to be effective for obsessive-compulsive disorder. This is the concept of drug repositioning, also known as the development of new drug indications, which refers to the process of analyzing the mechanism of action of drugs that have been approved for marketing or entered the clinical stage to find new uses beyond their original indications. This strategy can effectively shorten the cycle of new drug development, reduce R&D costs, and reduce safety risks.

[0003] With the rapid development of the information age, a large amount of biomedical data has been rapidly accumulated, including genomic data, transcriptome data, etc., which provides research support for drug repositioning. In recent years, the rise of deep learning technology has significantly promoted the processing of biomedical data. Through advanced algorithms such as deep neural networks, researchers can more effectively mine potential patterns and features in complex data.

[0004] At present, many studies at home and abroad have conducted effective research in the field of drug repositioning based on a large amount of biomedical data. By analyzing the similarity of molecular structure or biochemical characteristics between drugs and targets to predict the potential application of drugs in different diseases, the core is to assume that drugs with similar molecular structures may also have similar therapeutic effects in biological mechanisms, which provides a simple and scalable strategy for drug repositioning, especially under the premise of known drug targets or characteristics, by calculating the similarity of molecular characteristics, it is expected to predict and verify new drug-disease pairings. Napolitano et al. integrated gene expression profiles and target features, combined multiple data sources through machine learning, and successfully predicted new uses for multiple drugs. However, the molecular similarity calculation method relies too much on the assumption of similarity of molecular structure, and in fact many potential therapeutic effects may not be discovered due to differences in molecular structure. In addition, this method is based on the static structural characteristics of molecules and it is difficult to capture the dynamic and complex mechanism of action of drugs in biological systems.

[0005] In order to overcome the limitations of molecular similarity calculation methods, more complex network biology methods have been proposed. By integrating multi-source biological information to construct a disease-drug multi-source association network, the potential relationship between drugs and diseases can be further revealed. Ren ZH et al. constructed a heterogeneous network through a simplified graph convolutional network, combined with adaptive information diffusion distance to dynamically adjust the transmission range of network information, and strengthened the representation of drug-disease associations to achieve satisfactory results in drug repositioning tasks. Although the method based on multi-source association networks has shown significant performance, the complexity of its network also brings limitations that cannot be ignored. For example, the sources of data are diverse and the quality is uneven. The diversity of different types of nodes and relationships makes network representation learning complicated. Therefore, researchers explore the use of simpler network structures to focus on the direct association between drugs and diseases. Sun X et al. extracted partner-specific subgraphs of target drug-disease pairs in direct association networks and integrated multi-scale layer information using layer attention mechanisms. Jin S et al. generated multiple views of nodes in direct association networks through data augmentation methods such as node random inactivation, edge random inactivation, and random walks, and used graph convolutional networks to learn node embedding. At the same time, the learning process was further optimized through contrastive learning and adaptive denoising training. However, the above methods always deploy feature encoders in separate networks and cannot effectively guarantee the integrity of heterogeneous networks. Summary of the invention

[0006] In view of the above problems existing in the prior art, the present invention proposes a drug repositioning method based on a multi-level cross-domain relationship network, which has a reasonable design, solves the shortcomings of the prior art, and has good effects.

[0007] A drug repositioning method based on a multi-level cross-domain relationship network comprises the following steps: S1. Build a multi-level cross-domain relationship network , including drug-disease association native network, drug association network and disease association network; S2. Design a cross-domain fusion encoder to encode a multi-level cross-domain relationship network for drug repositioning tasks; S3, predicting the association between drugs and diseases using a multi-layer perceptron as a decoder; S4, a multi-level cross-domain relationship network, a cross-domain fusion encoder and a multi-layer perceptron are used to form a drug relocation model, and a classification task is constructed to train the model to obtain a trained drug relocation model; S5. Input the drug and disease to be tested into the trained drug repositioning model, and output the probability score of the test drug being effective for the test disease.

[0008] Furthermore, for the drug-disease association native network First, drug-disease information was collected through public databases, including PubChem, DrugCentral, and DrugBank, to build a direct relationship between drugs and diseases. Drugs for If the two diseases are cured, it is considered that the two There is a correlation, otherwise there is no correlation.

[0009] Furthermore, for the drug association network ,Will Defined as ,in yes The set of drug nodes in The initial embedding set for each drug, which is derived not only from the basic properties of the drug but also incorporates the DDI prediction score; for The association set between Chinese medicines; constructed using the k-nearest neighbor graph idea , that is, select the top k drugs from large to small according to the DDI prediction score to construct To achieve The purpose of association construction is expressed as: (1); in, For the Drugs and The association between drugs indicates that For the Drugs and The DDI scores between drugs , Indicates the amount of medicine, It includes The extended k-nearest neighbor set of the drug is constructed based on the k-nearest neighbor graph idea and DDI prediction score.

[0010] Furthermore, for the disease association network , based on the directed acyclic graph structure of MeSH, the semantic information of each disease is represented. Each disease is mapped into a DAG. Defined as , for The disease collection in is the semantic similarity of different disease DAGs, for The set of associations between diseases.

[0011] Furthermore, for the cross-domain fusion encoder, Defined as ,in, is The collection of drug nodes in is The set of disease nodes in Indicates the number of diseases, is The association set between Chinese medicine and disease, if Drugs for If the disease is effective, ,otherwise , ; As the input of the cross-domain fusion encoder, its initial embedding is defined as The entity index in is mapped to a one-hot index that can be learned to embed and and Entity association mapping in ; The information aggregation process of the cross-domain fusion encoder is shown in the formula: (2); in, express The initial input to the network, Represents the adjacency relationship between nodes in the network. is a learnable weight matrix, It is a complete convolution process, including two forms, correlation domain convolution and collaborative domain convolution , is the node type mapping function, for Middle nodes, For the nodes, ; Get the Drugs and The diseases are Embedding in and , where the first Drug introduction The specific information transmission process is: First, The initial association information of drug nodes is It has been provided in the formula (2). , the expression is: (3); (4); (5); in, For the The pseudo-correlation features corresponding to each drug, For the The real correlation features corresponding to each drug, For the The associated domain features corresponding to each drug, is the tanh activation function, and are all learnable weight matrices, is the element-wise multiplication aggregation function; For the The pseudo-correlation features corresponding to each drug, , For the The real correlation features corresponding to each drug, , and They are With Each drug has directly related neighbors and non-directly related neighbors; Then, the node features of the drug association network are obtained by Update, the expression is: (6); (7); in, For the The drug collaboration domain features corresponding to each drug, Yes The depth of the receptive field, It is Drugs in The neighbor set obtained according to the k-nearest domain, It is Drugs in The receptive field depth is The learnable weight matrix at It is Drugs in After Feature embedding after layer perception, After the update The insertion of a drug, is the Hadamard product aggregation function; Next, is fed into formula (2) for deeper exploration. Through deep feature encoding, not only The direct relationship between Chinese medicine and disease will also and Incorporate collaborative information into the encoding process; Finally, in the multi-layer architecture of the graph neural network, a layer-by-layer weighted fusion method is adopted to accumulate the features of each layer into the final output according to decreasing weights. The expression is: (8); in, It is Drugs in The final embedding obtained in For the The drug in The features obtained by the encoder are fused across layers. is the receptive domain depth of the cross-domain fusion encoder.

[0012] Further, in said S3, and Assigned to The corresponding drug node and disease node attributes are obtained by The update operation of the edge attribute integrates the features between drugs and diseases to generate the edge features between the target drug and disease. The expression is: (9); in, It is a feature integration function, which is used to integrate drug and disease features into network associations; Subsequently, the edge features are input into the multilayer perceptron to obtain high-order features layer by layer, and finally generate a prediction score for whether there is a therapeutic effect relationship between the target drug and the disease, expressed as: (10); in, is the prediction score made by the model, and is a learnable weight matrix, and is the bias parameter, is the ReLu activation function.

[0013] Furthermore, in S4, during the model training process, the drug repositioning problem is modeled as a binary classification task, that is, predicting whether the drug is effective for the disease. In order to optimize the model performance, the binary cross entropy is used as the loss function of the classification task, and the expression is: (11); in, is the true label.

[0014] Beneficial technical effects brought by the present invention: By constructing a multi-level cross-domain relationship network and integrating a variety of network architectures, the present invention can deeply mine more diverse semantic associations, thereby comprehensively improving the network's expression and reasoning capabilities. In order to overcome the limitations of traditional similarity-based drug association networks, the concept of DDI is introduced to more fully reveal the deep association between drugs at the biological level. A cross-domain fusion encoder is proposed, which significantly improves the accuracy of network coding by comprehensively encoding multi-source heterogeneous data. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 Schematic diagram of the traditional dual network architecture.

[0016] Figure 2 Schematic diagram of the multi-level cross-domain relationship network architecture in the present invention.

[0017] Figure 3 It is a schematic diagram of the process of the cross-domain fusion encoder in the present invention.

[0018] Figure 4 Schematic diagram of the principle of correlation domain convolution in the present invention.

[0019] Figure 5 Schematic diagram of the principle of collaborative domain convolution in the present invention. DETAILED DESCRIPTION

[0020] The specific implementation of the present invention is further described below in conjunction with specific embodiments: A drug repositioning method based on a multi-level cross-domain relationship network comprises the following steps: S1. Build a multi-level cross-domain relationship network , including drug-disease association native network, drug association network and disease association network; The traditional dual network architecture usually associates the drug-disease association with the original network As an independent architecture, the entity association network, namely the drug association network Disease-associated networks , modeled as separate architectures, such as Figure 1 Although it can capture the direct relationship between drugs and diseases and their internal relationships to a certain extent, the lack of deep interaction between the networks often leads to insufficient mining capabilities of potential relationships. To this end, the present invention proposes a multi-level cross-domain relationship network, such as Figure 2 As shown, while retaining the direct connection between network nodes, it further integrates Information about drugs and The inter-disease information of the original network is enriched, so that it can not only effectively express the direct relationship between drugs and diseases, but also fully explore the internal relationship between drugs and diseases to enhance the overall network, significantly improve the mapping ability of potential drug-disease associations, and achieve a high degree of integration of network structure.

[0021] For drug-disease association native network First, drug-disease information was collected through public databases, including PubChem, DrugCentral, and DrugBank, to build a direct relationship between drugs and diseases. Drugs for If the two diseases are cured, it is considered that the two There is a correlation, otherwise there is no correlation.

[0022] For drug association networks In existing methods, the construction of drug association networks is usually based on the similarities between drugs, such as using the similarities in structure, function or chemical properties. However, this method has significant limitations: the structural similarity of drugs does not always reflect the relevance of their mechanism of action in actual treatment. Certain structurally similar drugs may show significant differences in the activity of specific targets, and may even produce completely different therapeutic effects in disease treatment. Therefore, relying solely on similarity may cause the construction of drug association networks to lack certain biological significance, affecting the accuracy and reliability of entity associations. In order to solve this problem, the present invention collects a large amount of drug pair data from the DrugBank database and pre-trains a drug interaction prediction model as The discriminator of the interaction strength between drug pairs in the , and then the prediction scores between each drug pair are counted This method is different from the traditional strategy of building association networks based on drug similarities (such as structure or chemical properties). Instead, it provides a more biologically meaningful basis for network construction by predicting the actual functional associations between drugs.

[0023] Will Defined as ,in yes The set of drug nodes in The initial embedding set for each drug is derived from not only the basic properties of the drug, but also the DDI prediction score, which can more comprehensively reflect the association between drugs. For the DDI prediction score, a drug interaction prediction model is first built using drug structure and drug fingerprint, and then a large amount of drug pair data is collected from the DrugBank database for pre-training. Discriminator for DDI prediction score; for The association set between Chinese medicines; constructed using the k-nearest neighbor graph idea , that is, select the top k drug entities from large to small according to the DDI prediction score to construct To achieve The purpose of association construction is expressed as: (1); in, For the Drugs and The association between drugs indicates that For the Drugs and The DDI scores between drugs , Indicates the amount of medicine, It includes The extended k-nearest neighbor set of the drug is constructed based on the k-nearest neighbor graph idea and DDI prediction score.

[0024] For disease association networks , the directed acyclic graph structure based on MeSH (Medical Subject Headings) represents the semantic information of each disease. As a hierarchical medical subject vocabulary, MeSH can effectively reveal the hierarchy and correlation between diseases. Each disease is mapped to a DAG, where the nodes represent the subject terms of the disease and its upper-level concepts, and the edges represent the parent-child relationship between different levels. By calculating the semantic similarity of two disease DAGs, the degree of association between diseases can be quantified. The calculation of the similarity score is based on the co-occurrence information and hierarchical structure between nodes, which can comprehensively consider the specific characteristics of the disease and the higher-level semantic associations, and is used for The construction of Defined as , for The disease collection in is the semantic similarity of different disease DAGs, for The set of associations between diseases.

[0025] Build through differentiated strategies and , thereby more accurately portraying the complex relationship between drugs and diseases without relying on similarity assumptions, and being closer to the actual performance of drugs in disease treatment. It integrates drug networks based on functional associations and disease networks based on semantic associations, breaking through the limitations of traditional methods and achieving a comprehensive and accurate portrayal of the relationship between drugs and diseases.

[0026] Traditional dual network architecture usually As a standalone architecture, and Although modeling as a separate architecture can capture the direct relationship between drugs and diseases and their internal relationships to a certain extent, the lack of deep interaction between the networks often leads to insufficient mining of potential relationships. Each drug node and disease node retains its direct relationship and further integrates the Drug interaction information and The disease similarity information in the network can not only effectively express the direct relationship between drugs and diseases, but also fully explore the internal relationship between drugs and diseases to enhance the overall network, significantly improve the mapping ability of potential drug-disease associations, and achieve a high degree of integration of network structure.

[0027] S2. Design a cross-domain fusion encoder, encoding for drug repositioning missions; For cross-domain fusion encoders, such as Figure 3-5 As shown, Defined as ,in, is The collection of drug nodes in is The set of disease nodes in Indicates the number of diseases, is The association set between Chinese medicine and disease, if Drugs for If the disease is effective, ,otherwise , ; As the input of the cross-domain fusion encoder, its initial embedding is defined as The entity index in is mapped to a one-hot index that can be learned to embed and and Entity association mapping in ; The information aggregation process of the cross-domain fusion encoder is shown in the formula: (2); in, express The initial input to the network, Represents the adjacency relationship between nodes in the network. is a learnable weight matrix, It is a complete convolution process, including two forms, correlation domain convolution and collaborative domain convolution , is the node type mapping function, for Middle nodes, For the nodes, ; Get the Drugs and The diseases are Embedding in and , where the first Drug introduction The specific information transmission process is: First, The initial association information of drug nodes is It has been provided in the formula (2). , the expression is: (3); (4); (5); in, For the The pseudo-correlation features corresponding to each drug, For the The real correlation features corresponding to each drug, For the The associated domain features corresponding to each drug, is the tanh activation function, and are all learnable weight matrices, is the element-wise multiplication aggregation function, For the The pseudo-correlation features corresponding to each drug, , For the The real correlation features corresponding to each drug, , and They are With The purpose is to comprehensively capture the relationship between drug nodes in the basic network by aggregating the neighbor features of the two types of edge relationships. The diverse correlation information in provides an initial understanding for subsequent deeper feature representation; Then, the node features of the drug association network are obtained by Update, the expression is: (6); (7); in, For the The drug collaboration domain features corresponding to each drug, Yes The depth of the receptive field, It is Drugs in The neighbor set obtained according to the k-nearest domain, It is Drugs in The receptive field depth is The learnable weight matrix at It is Drugs in After Feature embedding after layer perception, After the update The insertion of a drug, is the Hadamard product aggregation function; Next, is fed into formula (2) for deeper exploration. Through deep feature encoding, not only The direct relationship between Chinese medicine and disease will also and The collaborative information in the is integrated into the encoding process, which enhances the potential relationship mapping ability between drug and disease nodes; Finally, in the multi-layer architecture of the graph neural network, the feature representations of different layers capture information at different levels: the shallower layers focus on the low-order features of the local neighborhood, while the deeper layers capture more global high-order features. However, over-reliance on the features of the last layer may cause the model to be biased towards high-order features and ignore the fine-grained information of low-order features. In addition, deep features may become unreliable due to vanishing gradients or excessive smoothing. Therefore, the present invention adopts a layer-by-layer weighted fusion method to accumulate the features of each layer into the final output according to decreasing weights, expressed as: (8); in, It is Drugs in The final embedding obtained in For the The drug in The features obtained by the encoder are fused across layers. is the receptive domain depth of the cross-domain fusion encoder.

[0028] Among them, the use of Disease Introduction The specific information transmission process is: First, The initial association information for each disease is It has been provided in the formula (2). , the expression is: (9); (10); (11); in, For the The pseudo-correlation features corresponding to each disease, For the The real correlation features corresponding to each disease, For the The associated domain features corresponding to each disease, and They are With The directly related neighbors and the indirect related neighbors corresponding to each disease, and are all learnable weight matrices; For the The pseudo-correlation features corresponding to each disease, For the The real correlation features corresponding to each disease; Then, the node features of the disease association network are obtained by Update, the expression is: (12); (13); in, For the Disease collaboration domain features corresponding to each disease, Yes The depth of the receptive field, It is The corresponding disease The neighbor set obtained according to the k-nearest domain, It is The corresponding disease The receptive field depth is The learnable weight matrix at It is Diseases in After Feature embedding after layer perception, After the update The embedding of disease is the Hadamard product aggregation function, is the associated domain feature; Finally, in the multi-layer architecture of the cross-domain fusion encoder, a layer-by-layer weighted fusion method is adopted to accumulate the features of each layer to the final output according to decreasing weights, and the expression is: (14); in, It is Diseases in The final embedding obtained in For the The disease is The features obtained by the encoder are fused across layers.

[0029] S3, predicting the association between drugs and diseases using a multi-layer perceptron as a decoder; and Assigned to The corresponding drug node and disease node attributes are obtained by The update operation of the edge attribute integrates the features between drugs and diseases to generate the edge features between the target drug and disease. The expression is: (15); in, It is a feature integration function, which is used to integrate drug and disease features into network associations; Subsequently, the edge features are input into the multilayer perceptron to obtain high-order features layer by layer, and finally generate a prediction score for whether there is a therapeutic effect relationship between the target drug and the disease, expressed as: (16); in, is the prediction score made by the model, and is a learnable weight matrix, and is the bias parameter, is the ReLu activation function.

[0030] S4, a multi-level cross-domain relationship network, a cross-domain fusion encoder and a multi-layer perceptron are used to form a drug relocation model, and a classification task is constructed to train the model to obtain a trained drug relocation model; There are many drug repositioning datasets available for research, such as Gdataset, Cdataset and LRSSL. Each dataset covers drug and disease information collected from multiple reliable databases. The known efficacy relationship is randomly divided into 10 subsets of equal size, and each subset is used as a test set in turn, and the remaining subsets are used as training sets. The whole process is repeated 10 times.

[0031] During the model training process, the drug repositioning problem is modeled as a binary classification task, that is, predicting whether the drug is effective for the disease. In order to optimize the model performance, binary cross entropy is used as the loss function of the classification task, expressed as: (17); in, is the true label.

[0032] S5. Input the drug and disease to be tested into the trained drug repositioning model, and output the probability score of the test drug being effective for the test disease.

[0033] Of course, the above description is not a limitation of the present invention, and the present invention is not limited to the above examples. Changes, modifications, additions or substitutions made by technicians in this technical field within the essential scope of the present invention should also fall within the protection scope of the present invention.

Claims

1. A drug repositioning method based on a multi-level cross-domain relationship network, characterized in that: The following steps are involved: S1. Build a multi-level cross-domain relationship network , including drug-disease association native network, drug association network and disease association network; S2. Design a cross-domain fusion encoder to encode a multi-level cross-domain relationship network for drug repositioning tasks; S3, predicting the association between drugs and diseases using a multi-layer perceptron as a decoder; S4, a multi-level cross-domain relationship network, a cross-domain fusion encoder and a multi-layer perceptron are used to form a drug relocation model, and a classification task is constructed to train the model to obtain a trained drug relocation model; S5. Input the drug and disease to be tested into the trained drug repositioning model, and output the probability score of the test drug being effective for the test disease.

2. A drug repositioning method based on a multi-level cross-domain relationship network according to claim 1, characterized in that: For drug-disease association native network First, drug-disease information was collected through public databases, including PubChem, DrugCentral, and DrugBank, to build a direct relationship between drugs and diseases. Drugs for If the two diseases are cured, it is considered that the two There is a correlation, otherwise there is no correlation.

3. A drug repositioning method based on a multi-level cross-domain relationship network according to claim 2, characterized in that: For drug association networks ,Will Defined as ,in yes The set of drug nodes in The initial embedding set for each drug, which is derived not only from the basic properties of the drug but also incorporates the DDI prediction score; for The association set between Chinese medicines; constructed using the k-nearest neighbor graph idea , that is, select the top k drugs from large to small according to the DDI prediction score to construct To achieve The purpose of association construction is expressed as: (1)； in, For the Drugs and The association between drugs indicates that For the Drugs and The DDI scores between drugs , Indicates the amount of medicine, It includes The extended k-nearest neighbor set of the drug is constructed based on the k-nearest neighbor graph idea and DDI prediction score.

4. A drug repositioning method based on a multi-level cross-domain relationship network according to claim 3, characterized in that: For disease association networks , based on the directed acyclic graph structure of MeSH, the semantic information of each disease is represented. Each disease is mapped into a DAG. Defined as , for The disease collection in is the semantic similarity of different disease DAGs, for The set of associations between diseases.

5. A drug repositioning method based on a multi-level cross-domain relationship network according to claim 4, characterized in that: For the cross-domain fusion encoder, Defined as ,in, is The collection of drug nodes in is The set of disease nodes in Indicates the number of diseases, is The association set between Chinese medicine and disease, if Drugs for If the disease is effective, ,otherwise , ; As the input of the cross-domain fusion encoder, its initial embedding is defined as The entity index in is mapped to a one-hot index that can be learned to embed and and Entity association mapping in ; The information aggregation process of the cross-domain fusion encoder is shown in the formula: (2)； in, express The initial input to the network, Represents the adjacency relationship between nodes in the network. is a learnable weight matrix, It is a complete convolution process, including two forms, correlation domain convolution and collaborative domain convolution , is the node type mapping function, for Middle nodes, For the nodes, ; Get the Drugs and The diseases are Embedding in and , where the first Drug introduction The specific information transmission process is: First, The initial association information of drug nodes is It has been provided in the formula (2). , the expression is: (3)； (4)； (5)； in, For the The pseudo-correlation features corresponding to each drug, For the The real correlation features corresponding to each drug, For the The associated domain features corresponding to each drug, is the tanh activation function, and are all learnable weight matrices, is the element-wise multiplication aggregation function; For the The pseudo-correlation features corresponding to each drug, , For the The real correlation features corresponding to each drug, , and They are With Each drug has directly related neighbors and non-directly related neighbors; Then, the node features of the drug association network are obtained by Update, the expression is: (6)； (7)； in, For the The drug collaboration domain features corresponding to each drug, Yes The depth of the receptive field, It is Drugs in The neighbor set obtained according to the k-nearest domain, It is Drugs in The receptive field depth is The learnable weight matrix at It is Drugs in After Feature embedding after layer perception, After the update The insertion of a drug, is the Hadamard product aggregation function; Next, is fed into formula (2) for deeper exploration. Through deep feature encoding, not only The direct relationship between Chinese medicine and disease will also and Incorporate collaborative information into the encoding process; Finally, in the multi-layer architecture of the graph neural network, a layer-by-layer weighted fusion method is adopted to accumulate the features of each layer into the final output according to decreasing weights. The expression is: (8)； in, It is Drugs in The final embedding obtained in For the The drug in The features obtained by the encoder are fused across layers. is the receptive domain depth of the cross-domain fusion encoder.

6. The drug repositioning method based on a multi-level cross-domain relationship network according to claim 5, characterized in that: In S3, and Assigned to The corresponding drug node and disease node attributes are obtained by The update operation of the edge attribute integrates the features between drugs and diseases to generate the edge features between the target drug and disease. The expression is: (9)； in, It is a feature integration function, which is used to integrate drug and disease features into network associations; Subsequently, the edge features are input into a multilayer perceptron to obtain high-order features layer by layer, and finally generate a prediction score for whether there is a therapeutic effect relationship between the target drug and the disease, expressed as: (10)； in, is the prediction score made by the model, and is a learnable weight matrix, and is the bias parameter, is the ReLu activation function.

7. The drug repositioning method based on a multi-level cross-domain relationship network according to claim 6, characterized in that: In S4, during the model training process, the drug repositioning problem is modeled as a binary classification task, that is, predicting whether the drug is effective for the disease. In order to optimize the model performance, binary cross entropy is used as the loss function of the classification task, and the expression is: (11)； in, is the true label.

Images