Hypergraph-based drug-target-disease interaction prediction method

Abstract

The invention discloses a drug-target-disease interaction prediction method based on a hypergraph. The method comprises the following steps: 1) establishing a semantic hypergraph G according to a binary relationship R between every two drugs, targets and diseases; according to the drug molecule fingerprints, the target sequences and the disease phenotypes, establishing feature similarity graphs of drug nodes, target nodes and disease nodes; 2) applying a hypergraph neural network on the semantic hypergraph G to obtain node representations corresponding to drugs, targets and diseases; respectively applying a graph convolutional network on the feature similarity graphs of the drugs, the targets and the diseases to obtain node representations corresponding to the drugs, the targets and the diseases; 3) fusing the node representations obtained in the step 2); 4) training a prediction model by using the fused node representation corresponding to each hyperedge obtained in the step 3); 5) generating node representations of a to-be-predicted drug a and a to-be-predicted disease c, and inputting the node representations into the trained prediction model for prediction to obtain a prediction probability that the drug a treats the disease c through the target b.

Description

technical field [0001] The invention belongs to the technical field of computer biological information network embedding and deep learning, and relates to a method for predicting drug-target-disease interaction relationships based on a hypergraph neural network. Background technique [0002] Modeling the behavior of drug-target-disease interactions is crucial in the early stages of drug discovery and holds great promise for precision medicine and personalized treatments. In the human metabolic system, many drugs interact with protein targets in cells and modulate target activity to alter biological pathways, promote healthy function and treat disease. Therefore, the mechanism of action of a drug can be better understood by using the closely related triple relationship of <drug, target, disease>. [0003] In recent years, the growth of data on the Internet and the continuous development of deep learning models have provided data support for exploring a more comprehensi...

AI Technical Summary

Problems solved by technology

However, most of the existing methods use the traditional graph structure (Graph), which maps a single data point to a node, and maps the connection or association between two points to an edge. However, in practical applications, this paired connection It is not enough to describe the complete data relationship, that is, this modeling method regards drug-disease and drug-target prediction as two independent tasks, focusing on the relationship between the two, ignoring the possible simultaneous play between the three entities. The dependence of action has limitations in representing complex data relationships among drugs, targets, and diseases

Images