Prediction method of protein interaction direction and regulation relationship

Abstract

The invention relates to a method for predicting protein interaction direction and regulatory relationship, which includes the following steps: performing homology modeling on the interacting proteins; performing homologous spatial structure of each interacting protein with each subunit of the protein complex Compare and overlay, select the stable complex structure, construct a protein interaction model with spatial structure resolution according to the stable complex structure, calculate the direction and frequency of the interaction of amino acid residues in the protein interaction interface, and generate the interaction protein amino acid residue Directed contact map: use convolutional neural network to extract target features of directed contact map, conduct recognition model training, obtain training model, and predict protein interaction direction and regulatory relationship. The prediction method of protein interaction direction and regulatory relationship is a prediction method based on deep learning, which can realize accurate and efficient prediction of protein interaction direction and regulatory relationship.

Description

Technical field [0001] The invention relates to the field of biotechnology, in particular to a method for predicting the direction and regulation relationship of protein interaction. Background technique [0002] The interaction between protein and protein is essential for almost all cellular processes. Most genes perform biological functions at the protein level by interacting with other proteins. [0003] At present, some high-throughput experimental methods have been developed to analyze the interaction relationship between proteins, which can deepen people's understanding of how proteins form functional complexes through interactions and participate in cellular responses to environmental stimuli. [0004] However, existing high-throughput protein interaction detection methods can only detect simple physical interactions (chemical bonds) between proteins, and lack the reflection of the directionality (upstream or downstream) of the interaction and the regulatory relationship (acti...

AI Technical Summary

Problems solved by technology

[0004] However, the existing high-throughput protein interaction detection methods can only detect simple physical interaction relationships (chemical bonds) between proteins, and lack the reflection of interaction directionality (upstream or downstream) and regulatory relationships (activation or inhibition). However, most of the interactions between proteins in real life systems have obvious biological functions such as signal transduction, activation, or inhibition.

Images