Target protein and micromolecule binding prediction method and system

Abstract

The invention provides a target protein and micromolecule binding prediction method and system. The target protein and micromolecule binding prediction method includes the steps: obtaining physicochemical characteristic data of a protein pocket to be bound and a small molecule to be queried; converting the physicochemical characteristic data of the protein pocket to be bound and the small moleculeto be queried into a pocket vector and a small molecule vector; and respectively inputting the pocket vector and the small molecule vector into a pre-set prediction model based on a deep neural network to obtain a prediction result of the binding probability. The target protein and micromolecule binding prediction method and system extract the active pocket part directly related to the interaction to represent protein, thus being beneficial to removing of non-related information and reduction of noise so as to improve accuracy. In addition, the target protein and micromolecule binding prediction method and system design a neural full-connected layer network model suitable for a learning vector to more easily retain more complete information, and can maintain the key information of the function of the small molecules of the protein by the vector while not depending on the protein small molecule complex conformation for laying the foundation for accurate prediction at high speed.

Description

technical field [0001] The invention relates to the field of computational biotechnology, in particular to a method and system for predicting the combination of a target protein and a small molecule. Background technique [0002] Protein is the basic functional unit of organisms, and small molecule drugs can affect disease-related physiological pathways by interacting with proteins. Designing small molecule drugs targeting disease targets has become one of the most important means of treating diseases. Experimental methods to predict or screen active candidate small molecules against proteins require a lot of money and take a long time. Therefore, a series of computer-aided methods have been developed to accelerate drug screening, among which protein-small molecule large-scale docking is widely used to find the best space interaction site and orientation of drug-protein, and finally determine the possible optimal complex by scoring function structure. However, researchers...

AI Technical Summary

Problems solved by technology

[0005] In view of the deficiencies in the prior art above, the purpose of the present invention is to provide users with a method and system for predicting the binding of target proteins and small molecules, which is used to overcome the problem of relying on accurate protein small molecule complex structures in the prior art. Models or machine learning algorithms for predicting interactions between proteins and small molecules cannot accurately predict the defects of binding results

Images