Protein structure prediction method based on geometric network

Abstract

The invention provides a protein structure prediction method based on a geometric network, and the method comprises the steps: S1, inputting an amino acid sequence and a PSSM matrix into a calculation unit, and obtaining a hidden state of the calculation unit; S2, inputting the hidden state of the calculation unit into a geometric unit, calculating a torsion angle of a given residue and a partially completed backbone generated by the upstream geometric unit by the geometric unit to obtain a new backbone of residue extension, and inputting the backbone into an adjacent downstream unit to obtain torsion angle information of protein, and finally, outputting a complete tertiary structure of the protein; and S3, calculating an error between the prediction structure and the experiment structure based on the dRMSD measurement.

Description

technical field [0001] The present invention relates to the fields of bioinformatics and computer applications, and further relates to the field of protein tertiary structure prediction, in particular, a geometric network-based protein structure prediction method. Background technique [0002] Protein structure prediction is a major problem in the field of life sciences. Almost every basic biological advance necessary for life has been made possible by proteins. Proteins are involved in creating cells and tissues and maintaining their shape, constituting enzymes that catalyze the chemical reactions needed to sustain life, acting as molecular factories, transporters and motors, serving as signals and receivers for cellular communication, and more. [0003] Proteins are made up of long chains of amino acids that perform myriad tasks by folding into precise 3D structures that control how they interact with other molecules. Proteins are made up of 20 different amino acids. Th...

AI Technical Summary

Problems solved by technology

But after nearly forty years of unremitting efforts, the puzzle remains unresolved

Images