Staged multi-strategy-based group conformation space sampling method

Abstract

The invention discloses a staged multi-strategy-based group conformation space sampling method. The method comprises the following steps: dividing a whole algorithm process into a plurality of stages under the framework of a differential evolution algorithm, setting a group of strategy pools for each stage, and when the algorithm achieves a certain stage, randomly selecting a strategy from the corresponding strategy pool to generate a new test conformation individual on the basis of a fragment assembling technology, so as to improve the conformation sampling ability and the convergence rate of the algorithm; introducing distance spectrum constraint in the conformation selection link, when the energy of the tested conformation is higher than a target conformation, comparing the distance differences of the tested conformation and of the target conformation, and if the distance difference of the test individual is relatively small, receiving the tested conformation at a certain probability, so as to guide the algorithm sampling to obtain a conformation with lower energy and a more reasonable structure and then improve the prediction precision of the algorithm.

Description

technical field [0001] The invention relates to the fields of bioinformatics and computer applications, in particular to a group conformation space sampling method based on multiple strategies in stages. Background technique [0002] There are many proteins (long chains formed by two kinds of amino acids) in biological cells. These macromolecules play an important role in living organisms and are essential for completing biological functions. Protein molecules reflect the important relationship between their structure and function at the molecular level, and different proteins play different roles in organisms. The spatial structure of a protein often determines its function. Therefore, the prediction of protein structure is crucial for the design of new proteins, drug design, prediction of protein stability, and modeling of protein-protein interactions. [0003] The structure of protein is generally divided into four levels: primary structure (amino acid sequence), seconda...

AI Technical Summary

Problems solved by technology

[0007] In order to overcome the deficiencies in sampling ability and convergence speed of the existing population conformation space sampling method, the present invention provides a population conformation space sampling method based on staged multi-strategy that improves sampling ability, convergence speed, and prediction accuracy

Images