Graph-Transform-based drug target interaction prediction method research

Abstract

The invention discloses a graph-Transform-based drug target interaction prediction method research, which aims to predict whether drug target pairs interact or not, and the method mainly comprises the steps of node feature processing, graph structure processing, model definition, model training and model evaluation. Rapid and accurate discovery of drugs capable of effectively treating diseases is very important for social development, but based on prediction among drug targets, the relationship among the drug targets cannot be completely obtained, information of some surrounding nodes is ignored, some current models cannot well extract features among global nodes, and based on the above consideration, the invention provides a drug target interaction prediction method based on a graph Transform. Whether drug target pairs interact or not can be automatically recognized.

Description

technical field [0001] The invention belongs to the field of biochemistry, and is a research on a drug target interaction prediction method based on graph Transformer. Background technique [0002] Today, tens of thousands of diseases are known to threaten human health, and new diseases are added every year. Therefore, it is very important for the development of society to quickly and accurately discover drugs that can effectively treat diseases. Drug-target prediction is to identify new ligands for new drugs and protein targets by identifying the interaction between drug compounds and protein targets, which can greatly reduce the cost and time of experiments. Traditional DTIs are roughly divided into two categories: one is based on molecular docking methods, and the other is based on ligand methods. These methods have obvious shortcomings. For example, the molecular docking method depends on the three-dimensional structure of the target protein. In reality, the three-dimen...

AI Technical Summary

Problems solved by technology

However, most computational methods do not take into account the graph Transformer

The Transformer model fully realizes the explicit information interaction between different components in Attention, but it is only verified in the sequence model and has not been applied to the graph structure.

Images