New coronavirus target prediction and drug discovery method based on graph representation learning

Abstract

The invention belongs to the field of bioinformatics, and discloses a new coronavirus target prediction and drug discovery method based on graph representation learning, which comprises the followingsteps of: constructing a heterogeneous network comprising action relationships among drugs, targets, side effects and diseases, and integrating topological structure information of the heterogeneous network and various fields of the heterogeneous network by using a graph convolutional neural network, combining an attention mechanism to reflect the importance degree of different types of neighborhood information to nodes, further learning the feature representation of the nodes, establishing a topological reconstruction process, and finally forcibly extracting the feature representation of drugs and targets through the topological reconstruction process to obtain a relocation network of which the edge weight represents the action intensity of the relationship. The target of the novel coronavirus is predicted according to the potential relationship of individual drugs for inhibiting the novel coronavirus in the relocation network, and the drugs for possibly inhibiting the novel coronavirus are picked out. In this way, the drugs capable of inhibiting the novel coronavirus can be effectively screened out, drug research and development are accelerated, and very important application andpopularization value is achieved.

Description

technical field [0001] The invention relates to a new coronavirus target prediction and drug discovery method based on graph representation learning, and belongs to the field of bioinformatics. Background technique [0002] New Coronary Pneumonia (COVID-19), caused by a novel coronavirus (2019-nCoV, SARS-CoV-2), urgently needs to discover or develop more drugs that can inhibit this virus. Drug repositioning can not only save a lot of drug design and screening costs in the early stage of drug development, but also significantly reduce the risk in the later stage of research and development because the pharmacokinetic properties and toxicity of the used drugs have been thoroughly studied. Using computer prediction methods to study the correlation of drug targets and narrow the search space of candidate experimental drugs can provide a reference for drug discovery and repositioning and reduce the corresponding time investment and cost consumption. [0003] Many efforts in rece...

AI Technical Summary

Problems solved by technology

However, when it aggregates the information of different neighbor nodes, it treats them equally, and there are certain limitations.

Images