Inductive Traditional Chinese Medicine Target Relationship Discovery Method and System Based on Multi-Scale HGT

Abstract

The present invention proposes an inductive traditional Chinese medicine target relationship discovery method and system based on multi-scale HGT. The method includes: constructing a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and performing node initialization operations on the heterogeneous information graph; inputting the initialized node features into the topological robust local attention for local representation; inputting the initialized node features into the global attention for global node representation; inputting the local representation of the nodes and the global representation of the nodes into the feature fusion module to perform concatenation operations and feature fusion in sequence; inputting the final feature representation into the prediction model for prediction to obtain the prediction result. The present invention comprehensively characterizes the rich semantics and multiple relationships among traditional Chinese medicine, targets, molecules, and diseases extracted from the traditional Chinese medicine network pharmacology database by constructing a new traditional Chinese medicine-target heterogeneous graph.

Description

Technical Field

[0001] The present invention relates to the field of traditional Chinese medicine (TCM)-target relationship prediction, and particularly to an inductive TCM-target relationship discovery method and system based on multi-scale heterogeneous graph transformer (HGT). Background Art

[0002] Predicting traditional Chinese medicine-target interactions (HTIs) plays a crucial role in revealing the pharmacological mechanisms of traditional Chinese medicine (TCM) and accelerating new drug development. Although deep learning methods have shown great potential in this field, existing methods still face some challenges, including difficulties in modeling the complex relationships between TCM entities, the problem of hidden biases, and poor generalization ability for unknown data. These challenges hinder the model's ability to effectively capture the complex interactions between biomedical entities and lead to poor performance in predicting newly discovered elements such as molecular or protein targets. Summary of the Invention

[0003] In view of the above situation, the main object of the present invention is to propose an inductive TCM-target relationship discovery method and system based on multi-scale HGT to solve the above technical problems.

[0004] The present invention proposes an inductive TCM-target relationship discovery method based on multi-scale HGT, and the method includes the following steps:

[0005] Step 1: Construct local attention based on a topological robust local attention mechanism, construct global attention based on a multi-relational global cross-attention mechanism, construct a feature fusion module based on a feature fusion mechanism, and the local attention, global attention, feature fusion module, and prediction module constitute a prediction model;

[0006] Among them, the global attention includes a feed-forward network, and the prediction module includes a linear layer;

[0007] Step 2: Construct a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and perform node initialization operations on the heterogeneous information graph to obtain node initialization features;

[0008] Step 3: Input the node initialization features into the topological robust local attention for local representation to obtain the local representation of the nodes;

[0009] Step 4: Input the node initialization features into the global attention for global node representation to obtain the global representation of the nodes;

[0010] Step 5: Input the local representation of the nodes and the global representation of the nodes into the feature fusion module for concatenation operation and feature fusion processing in sequence to obtain the final feature representation;

[0011] Step 6: Input the final feature representation into the linear layer for prediction to obtain a predicted value;

[0012] Construct a binary cross - entropy loss based on the predicted value, and use the binary cross - entropy loss to optimize the prediction model to obtain the optimized prediction model;

[0013] Use the optimized prediction model to obtain the final prediction result.

[0014] The present invention also proposes an inductive traditional Chinese medicine target relationship discovery system based on multi - scale HGT. The system includes:

[0015] A construction module, used for:

[0016] Construct local attention based on the topological robust local attention mechanism, construct global attention based on the multi - relation global cross - attention mechanism, construct a feature fusion module based on the feature fusion mechanism. The local attention, global attention, feature fusion module and prediction module constitute the prediction model;

[0017] Among them, the global attention includes a feed - forward network, and the prediction module includes a linear layer;

[0018] A data processing module, used for:

[0019] Construct a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and perform node initialization operations on the heterogeneous information graph to obtain node initialization features;

[0020] A feature extraction module, used for:

[0021] Input the node initialization features into the topological robust local attention for local representation to obtain the local representation of the nodes;

[0022] Input the node initialization features into the global attention for global node representation to obtain the global representation of the nodes;

[0023] A feature processing module, used for:

[0024] Input the local representation of the nodes and the global representation of the nodes into the feature fusion module for concatenation operations and feature fusion processing in sequence to obtain the final feature representation;

[0025] A relationship prediction module, used for:

[0026] Input the final feature representation into the linear layer for prediction to obtain the predicted value;

[0027] Construct a binary cross - entropy loss based on the predicted value, and use the binary cross - entropy loss to optimize the prediction model to obtain the optimized prediction model;

[0028] Use the optimized prediction model to obtain the final prediction result.

[0029] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0030] 1. The present invention comprehensively characterizes the rich semantics and multiple relationships among traditional Chinese medicines, targets, molecules, and diseases extracted from the traditional Chinese medicine network pharmacology database by constructing a new traditional Chinese medicine-target heterogeneous graph;

[0031] 2. The present invention processes diverse information on the traditional Chinese medicine-target heterogeneous graph through a graph neural network model, helping the model to more fully understand the meaning conveyed on the traditional Chinese medicine-target heterogeneous graph and enabling it to more accurately obtain key information from the complex traditional Chinese medicine relationship graph;

[0032] 3. The present invention enhances feature extraction by proposing a heterogeneous graph transformer based on multi-scale relationships, which can capture different aspects of entities on the traditional Chinese medicine-target heterogeneous graph during the learning process. This mechanism effectively captures remote node dependencies, reduces the impact of hidden biases, and promotes efficient traditional Chinese medicine target prediction in inductive learning scenarios;

[0033] 4. The present invention fuses the local features and global features of nodes by means of a multi-level attention mechanism and a fusion mechanism, providing an effective feature representation for the generation of traditional Chinese medicine-target link prediction in the final relationship prediction layer.

[0034] Additional aspects and advantages of the present invention will be partially given in the following description, partially become apparent from the following description, or be understood through the embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] Figure 1 It is a flowchart of the steps of the inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT proposed by the present invention.

[0036] Figure 2 It is a process diagram of obtaining node features based on local attention of the inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT proposed by the present invention.

[0037] Figure 3 It is a process diagram of obtaining node features based on global attention of the inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT proposed by the present invention.

[0038] Figure 4 It is a process diagram of feature fusion and relationship prediction of the inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT proposed by the present invention.

[0039] Figure 5 It is a schematic structural diagram of the inductive traditional Chinese medicine target relationship discovery system based on multi-scale HGT proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0040] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings, where the same or similar reference numerals throughout denote the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention and should not be construed as a limitation of the present invention.

[0041] These and other aspects of the embodiments of the present invention will be clear with reference to the following description and drawings. In these descriptions and drawings, some specific embodiments of the embodiments of the present invention are specifically disclosed as some ways to implement the principles of the embodiments of the present invention, but it should be understood that the scope of the embodiments of the present invention is not limited thereto.

[0042] Please refer to Figure 1 , an inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT is proposed in the embodiments of the present invention. The method includes the following steps:

[0043] Step 1: Construct local attention based on the topological robust local attention mechanism, construct global attention based on the multi-relation global cross-attention mechanism, and construct a feature fusion module based on the feature fusion mechanism. The local attention, global attention, and feature fusion module constitute a prediction model;

[0044] Among them, the global attention includes a feed-forward network, and the prediction module includes a linear layer.

[0045] Step 2: Construct a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and perform node initialization operations on the heterogeneous information graph to obtain node initialization features.

[0046] Step 3: Input the node initialization features into the topological robust local attention for local representation to obtain the local representation of the nodes.

[0047] Please refer to Figure 2 , in Step 3, input the node initialization features into the topological robust local attention, and perform local representation on the heterogeneous information graph to obtain the local representation of the nodes, which specifically includes the following steps:

[0048] Input the node initialization features into the topological robust local attention and perform a projection operation using the local weight matrix to obtain the features of the local attention nodes after projection;

[0049] Use an edge to connect two adjacent nodes as a node pair, project the node pair, and obtain the node representation of the node pair;

[0050] Perform attention calculation on the node representation of the node pair to obtain the attention value between the node pairs;

[0051] Aggregate operation is performed using the attention values between node pairs to obtain aggregated heterogeneous neighbor information. The aggregated heterogeneous neighbor information is used for local representation by combining the node representations of the node pairs with the attention values between the node pairs, so as to obtain the local representation of the nodes.

[0052] The initial node features are input into the topologically robust local attention for projection operation using the local weight matrix, and the features of the nodes after local attention projection are obtained. The corresponding relationship in the process is as follows:

[0053] ;

[0054] Among them, represents the features of node after projection in the local attention, represents the local weight matrix of node type ; represents the initial features of the nodes;

[0055] In the step of taking two adjacent nodes connected by an edge as a node pair and projecting the node pair to obtain the node representation of the node pair, the corresponding relationship in the process is as follows:

[0056] ;

[0057] Among them, represents the key representation of node , represents the query representation of node , represents the value representation of node , represents the parameter matrix used to project the input of node into the key space, represents the parameter matrix used to project the input of node into the query space, represents the parameter matrix used to project the input of node into the value space, represents the edge, represents the key vector, represents the query vector, represents the value vector, represents the linear layer, represents the function to obtain the relationship of edge ;

[0058] In the step of calculating the attention between the node representations of the node pairs to obtain the attention values between the node pairs, the corresponding relationship in the process is as follows:

[0059] ;

[0060] Among them, represents the node For the node attention value, represents a non-linear activation function, represents the row weight vector of a specific relationship, represents the weight matrix of a specific relationship, represents the concatenation operation;

[0061] It should be noted that maps the input value to the range of (0, 1).

[0062] In the step of using the attention value between node pairs for aggregation operation to obtain aggregated heterogeneous neighbor information, and performing local representation on the aggregated heterogeneous neighbor information by combining the node representations of node pairs with the attention value between node pairs to obtain the local representation of nodes, the relational expressions existing in the corresponding process are as follows:

[0063] ;

[0064] Among them, represents the local representation of the node , represents the local output weight matrix related to the node type , represents a linear layer, represents the node through the relationship neighbor set, represents the set of relationship types.

[0065] It should be noted that the relationship type refers to different types of edges between nodes.

[0066] Furthermore, on the constructed heterogeneous information graph, it aims to capture the semantic and topological relationships between different node types. To achieve this goal, a simple and effective Topological Robust Local Attention (TRLA) is used to aggregate local information from the source node to the target node to capture fine-grained structures.

[0067] Step 4: Input the node initialization features into the global attention for global node representation to obtain the global representation of the nodes.

[0068] Please refer to Figure 3 In step 4, inputting the node initialization features into the global attention for global node representation to obtain the global representation of the nodes specifically includes the following steps:

[0069] The initial features of the nodes are input into the global attention for projection operation to obtain the features of the nodes after projection in the global attention;

[0070] The features of the nodes after projection in the global attention are subjected to node partitioning operation to obtain a target node set and a source node set;

[0071] The target node set is used to construct a target node embedding sequence to obtain the constructed target node embedding sequence;

[0072] The source node set is used to construct a source node embedding sequence to obtain the constructed source node embedding sequence;

[0073] The target node embedding sequence and the source node embedding sequence are successively subjected to self-attention mechanism enhancement processing and feed-forward network processing to obtain an enhanced target node representation and an enhanced source node representation respectively;

[0074] The enhanced target node representation and the enhanced source node representation are subjected to cross-attention mechanism processing to obtain the global representation of the nodes.

[0075] The initial features of the nodes are input into the global attention for projection operation to obtain the features of the nodes after projection in the global attention. The corresponding relationship in the process is as follows:

[0076] ;

[0077] Among them, represents the features of the nodes after projection in the global attention, represents the node type of the global weight matrix;

[0078] In the step of subjecting the features of the nodes after projection in the global attention to node partitioning operation to obtain a target node set and a source node set, the corresponding relationship in the process is as follows:

[0079] ;

[0080] Among them, represents the target node set, represents all the node sets in the heterogeneous information graph, represents a single node, is used to obtain the category of the node, represents the source node set, is used to obtain the node and the node relationship category between;

[0081] In the step of constructing the target node embedding sequence by using the target node set to obtain the constructed target node embedding sequence, the relational expressions existing in the corresponding process are as follows:

[0082] ;

[0083] Among them, represents the embedding matrix of the target node, represents the first embedding of the target node, represents the second embedding of the target node, represents the th embedding of the target node, represents the transpose;

[0084] In the step of constructing the source node embedding sequence by using the source node set to obtain the constructed source node embedding sequence, the relational expressions existing in the corresponding process are as follows:

[0085] ;

[0086] Among them, represents the embedding matrix of the source node, represents the th embedding of the source node.

[0087] Performing self-attention mechanism enhancement processing and feed-forward network processing on the target node embedding sequence and the source node embedding sequence in sequence to obtain the enhanced target node representation and the enhanced source node representation respectively, the relational expressions existing in the corresponding process are as follows:

[0088] ;

[0089] Among them, represents the representation of the target node after self-attention enhancement, represents the representation of the source node after self-attention enhancement, represents the self-attention mechanism of the node, represents the feed-forward network;

[0090] In the step of performing cross-attention mechanism processing on the enhanced target node representation and the enhanced source node representation to obtain the global representation of the target node, the relational expressions existing in the corresponding process are as follows:

[0091] ;

[0092] Among them, represents the global representation matrix of the target node, represents the query matrix, represents the first weight matrix, represents the key matrix, represents the second weight matrix, represents the value matrix, represents the third weight matrix, represents the normalization function, represents the square root of the dimension of the embedding vector, represents the key matrix transpose of , represents the node global representation of, represents mapping the identifier of node to the specific row number in the matrix.

[0093] It should be noted that by jointly utilizing self-attention and cross-attention in the global graph, the global representations of all node types can be calculated.

[0094] Furthermore, on the constructed heterogeneous information graph, newly arrived nodes are usually isolated, which makes it difficult to learn comprehensive node representations through message passing strategies. For this reason, a new multi-relational global attention (MRGA) is proposed, which can automatically capture long-range dependencies in the entire heterogeneous information graph.

[0095] Step 5: Input the local representation and the global representation of the target node into the feature fusion module to perform concatenation operation and feature fusion processing in sequence to obtain the final feature representation.

[0096] Please refer to Figure 4 , in Step 5, input the local representation and the global representation of the node into the feature fusion module to perform concatenation operation and feature fusion processing in sequence to obtain the final feature representation, and the relational expressions existing in the corresponding process are as follows:

[0097] ;

[0098] Among them, represents the final feature representation of node obtained by concatenating the local representation of node and the global representation of node and then performing fusion calculation through a specific type of transformation matrix .

[0099] Step 6: Input the final feature representation into the linear layer for prediction to obtain the predicted value;

[0100] Construct a binary cross-entropy loss based on the predicted value, and use the binary cross-entropy loss to optimize the prediction model to obtain the optimized prediction model;

[0101] Use the optimized prediction model to obtain the final prediction result.

[0102] In step 6, a binary cross-entropy loss is constructed based on the predicted value, and the relationship in the corresponding process is as follows:

[0103] ;

[0104] Among them, represents the binary cross-entropy loss, represents the true label, represents the logarithmic function, represents the generated predicted interaction probability.

[0105] Furthermore, first take the traditional Chinese medicine node as the target node and process it through local attention and global attention in turn to obtain the local feature of the traditional Chinese medicine and the global feature of the traditional Chinese medicine. Then, fuse the obtained local feature of the traditional Chinese medicine and the global feature of the traditional Chinese medicine in the fusion layer of the feature fusion module to obtain the final representation of the traditional Chinese medicine node; in the same way, take the target node as the target node and go through the same process to obtain the local feature of the target and the global feature of the target respectively. Then, fuse the local feature of the target and the global feature of the target into the final traditional Chinese medicine feature and perform feature fusion in the fusion layer to obtain the final representation of the target node. Finally, input the final representation of the traditional Chinese medicine node and the final representation of the target node into the prediction module containing three fully connected layers for prediction.

[0106] Please refer to Figure 5 , the present invention also proposes an inductive traditional Chinese medicine target relationship discovery system based on multi-scale HGT. The system includes:

[0107] A construction module, used for:

[0108] Construct local attention based on the topological robust local attention mechanism, construct global attention based on the multi-relation global cross-attention mechanism, and construct a feature fusion module based on the feature fusion mechanism. The local attention, global attention, feature fusion module, and prediction module constitute a prediction model;

[0109] Among them, the global attention includes a feed-forward network, and the prediction module includes a linear layer;

[0110] A data processing module, used for:

[0111] Construct a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and perform node initialization operations on the heterogeneous information graph to obtain node initialization features;

[0112] A feature extraction module, used for:

[0113] Input the initialized features of the nodes into the topologically robust local attention for local representation to obtain the local representations of the nodes;

[0114] Input the initialized features of the nodes into the global attention for global node representation to obtain the global representations of the nodes;

[0115] A feature processing module for:

[0116] Input the local representations and global representations of the nodes into the feature fusion module to perform concatenation operations and feature fusion processing in sequence to obtain the final feature representation;

[0117] A relationship prediction module for:

[0118] Input the final feature representation into a linear layer for prediction to obtain a predicted value;

[0119] Construct a binary cross-entropy loss based on the predicted value, and use the binary cross-entropy loss to optimize the prediction model to obtain an optimized prediction model;

[0120] Use the optimized prediction model to obtain the final prediction result.

[0121] It should be understood that each part of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above embodiments, multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one or a combination of the following well-known technologies in the art can be used: discrete logic circuits with logic gate circuits for implementing logic functions on data signals, application-specific integrated circuits with appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field-programmable gate arrays (FPGAs), etc.

[0122] In the description of this specification, the description with reference to terms such as "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials, or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials, or characteristics described can be combined in a suitable manner in any one or more embodiments or examples.

[0123] The above-described embodiments merely represent several implementation manners of the present invention. The description thereof is relatively specific and detailed, but it should not be construed as a limitation to the scope of the patent of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all fall within the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims

1. An inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT, characterized in that The method includes the following steps: Step 1: Construct local attention based on the topological robust local attention mechanism, construct global attention based on the multi-relation global cross-attention mechanism, construct a feature fusion module based on the feature fusion mechanism. The local attention, global attention, feature fusion module, and prediction module constitute a prediction model; Among them, the global attention includes a feed-forward network, and the prediction module includes a linear layer; Step 2: Construct a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and perform node initialization operations on the heterogeneous information graph to obtain node initialization features; Step 3: Input the node initialization features into the topological robust local attention for local representation to obtain the local representation of the nodes; Step 4: Input the node initialization features into the global attention for global node representation to obtain the global representation of the nodes; Step 5: Input the local representation of the nodes and the global representation of the nodes into the feature fusion module to perform concatenation operations and feature fusion processing in sequence to obtain the final feature representation; Step 6: Input the final feature representation into the linear layer for prediction to obtain a predicted value; Construct a binary cross-entropy loss based on the predicted value, and use the binary cross-entropy loss to optimize the prediction model to obtain an optimized prediction model; Use the optimized prediction model to obtain the final prediction result; In the said Step 3, inputting the node initialization features into the topological robust local attention for local representation to obtain the local representation of the nodes specifically includes the following steps: Input the node initialization features into the topological robust local attention, and perform a projection operation using the local weight matrix to obtain the features of the locally attentioned nodes after projection; Take the edge connecting two adjacent nodes as a node pair, project the node pair to obtain the node representation of the node pair; Perform attention calculation on the node representation of the node pair to obtain the attention value between the node pairs; Use the attention value between the node pairs for an aggregation operation to obtain aggregated heterogeneous neighbor information, and perform local representation on the aggregated heterogeneous neighbor information through the node representation of the node pair and in combination with the attention value between the node pairs to obtain the local representation of the nodes; In the said Step 4, inputting the node initialization features into the global attention for global node representation to obtain the global representation of the nodes specifically includes the following steps: Input the node initialization features into the global attention for a projection operation to obtain the features of the nodes after projection in the global attention; Perform a node partitioning operation on the features of the nodes after projection in the global attention to obtain a target node set and a source node set; Use the target node set to construct a target node embedding sequence to obtain the constructed target node embedding sequence; Use the source node set to construct a source node embedding sequence to obtain the constructed source node embedding sequence; Perform self-attention mechanism enhancement processing and feed-forward network processing on the target node embedding sequence and the source node embedding sequence in sequence to obtain an enhanced target node representation and an enhanced source node representation respectively; Perform cross-attention mechanism processing on the enhanced target node representation and the enhanced source node representation to obtain the global representation of the nodes.

2. The inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT according to claim 1, wherein Input the initialized features of the nodes into the topological robust local attention, and use the local weight matrix for projection operation to obtain the features of the nodes after local attention projection. The relational expressions existing in the corresponding process are as follows: ； Among them, represents the projected feature of node in local attention, represents the local weight matrix of node type , and represents the initial feature of the node. In the step of taking the edges connecting two adjacent nodes as node pairs, projecting the node pairs, and obtaining the node representations of the node pairs, the relational expressions existing in the corresponding process are as follows: ； Among them, represents a node whose key represents represents a node whose query represents represents a node whose value represents represents a parameter matrix for projecting the input of node onto the key space represents a parameter matrix for projecting the input of node onto the query space represents a parameter matrix for projecting the input of node onto the value space represents an edge represents a key vector represents a query vector represents a value vector represents a linear layer represents a function for obtaining the relationship of edge ;​​​ In the step of calculating the attention of the node representations of the node pairs to obtain the attention values between the node pairs, the relational expressions existing in the corresponding process are as follows: ； Among them, represents a node The attention value of the node is represents a non-linear activation function, represents the row weight vector of a specific relationship, represents the weight matrix of a specific relationship, represents a concatenation operation; In the step of using the attention values between the node pairs for aggregation operation to obtain aggregated heterogeneous neighbor information, and performing local representation on the aggregated heterogeneous neighbor information through the node representations of the node pairs and combining the attention values between the node pairs to obtain the local representation of the nodes, the relational expressions existing in the corresponding process are as follows: ； Among them, represents the local representation of the node . represents the local output weight matrix related to the node type . represents the linear layer represents the node through the relationship neighbor set represents the set of relationship types 3. The inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT according to claim 2, wherein Input the initialized features of the nodes into the global attention for projection operation to obtain the features of the nodes after projection in the global attention. The relational expressions existing in the corresponding process are as follows: ； Among them, represents the feature after node projection in global attention, represents the node type of the global weight matrix; In the step of performing node partitioning operation on the features of the nodes after projection in the global attention to obtain the target node set and the source node set, the relational expressions existing in the corresponding process are as follows: ； Among them, represents the set of target nodes, represents the set of all nodes in the heterogeneous information graph, represents a single node, is used to obtain the category of the node, represents the set of source nodes, is used to obtain the node and the node the relationship category between; In the step of using the target node set to construct the target node embedding sequence to obtain the constructed target node embedding sequence, the relational expressions existing in the corresponding process are as follows: ； Among them, represents the embedding matrix of the target node, represents the first embedding of the target node, represents the second embedding of the target node, represents the -th embedding of the target node, represents the transpose; In the step of using the source node set to construct the source node embedding sequence to obtain the constructed source node embedding sequence, the relational expressions existing in the corresponding process are as follows: ； Among them, represents the embedding matrix of the source node, represents the th embedding of the source node.

4. The inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT according to claim 3, wherein Perform self-attention mechanism enhancement processing and feed-forward network processing on the target node embedding sequence and the source node embedding sequence in turn to obtain the enhanced target node representation and the enhanced source node representation respectively. The relational expressions existing in the corresponding process are as follows: ； Among them, represents the representation of the target node after self-attention enhancement, represents the representation of the source node after self-attention enhancement, represents the self-attention mechanism of the node, represents the feed-forward network; In the step of performing cross-attention mechanism processing on the enhanced target node representation and the enhanced source node representation to obtain the global representation of the nodes, the relational expressions existing in the corresponding process are as follows: ； Among them, represents the global representation matrix of the target node, represents the query matrix, represents the first weight matrix, represents the key matrix, represents the second weight matrix, represents the value matrix, represents the third weight matrix, represents the normalization function, represents the square root of the dimension of the embedding vector, represents the key matrix transpose of , represents the node global representation of, represents mapping the identifier of the node to the specific row number in the matrix.

5. The inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT according to claim 4, characterized in that In step 5, input the local representation and the global representation of the nodes into the feature fusion module to perform concatenation operation and feature fusion processing in turn to obtain the final feature representation. The relational expressions existing in the corresponding process are as follows: ； Among them, represents the local representation of node and the global representation of node are concatenated and then fused through a specific type of transformation matrix to calculate the final feature representation of node and then fused through a specific type of transformation matrix to calculate the final feature representation of node .

6. The inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT according to claim 5, wherein In step 6, construct the binary cross-entropy loss based on the predicted values. The relational expressions existing in the corresponding process are as follows: ； Among them, represents binary cross-entropy loss, represents the true label, represents the logarithmic function, represents the generated predicted interaction probability.

7. An inductive traditional Chinese medicine target relationship discovery system based on multi-scale HGT, characterized in that The system applies the inductive traditional Chinese medicine target relationship discovery method based on multi-scale HGT as described in any one of claims 1 to 6 above. The system includes: A construction module for: Constructing local attention based on the topological robust local attention mechanism, constructing global attention based on the multi-relation global cross-attention mechanism, constructing a feature fusion module based on the feature fusion mechanism. The local attention, global attention, feature fusion module and prediction module constitute a prediction model; Among them, the global attention includes a feed-forward network, and the prediction module includes a linear layer; A data processing module for: Constructing a heterogeneous information graph based on the original data of traditional Chinese medicine, drug molecules, targets, and diseases, and performing node initialization operation on the heterogeneous information graph to obtain the initialized features of the nodes; A feature extraction module for: Input the initialized features of the nodes into the topologically robust local attention for local representation to obtain the local representation of the nodes; Input the initialized features of the nodes into the global attention for global node representation to obtain the global representation of the nodes; A feature processing module, configured to: Input the local representation of the nodes and the global representation of the nodes into the feature fusion module to perform a concatenation operation and feature fusion processing in sequence to obtain the final feature representation; A relationship prediction module, configured to: Input the final feature representation into a linear layer for prediction to obtain a predicted value; Construct a binary cross-entropy loss based on the predicted value, and use the binary cross-entropy loss to optimize the prediction model to obtain an optimized prediction model; Use the optimized prediction model to obtain the final prediction result.

Images