Protein interaction site identification method

Abstract

The invention discloses a protein interaction site identification method which belongs to the field bioinformatics analysis. The method comprises the steps of acquiring protein chain data, performing preprocessing on the protein chain data, and dividing the preprocessed protein chain data to an interface residue and a non-interface residue; extracting a characteristic from a database, fusing the extracted characteristics for obtaining a data set, processing the unbalance of the data set, then dividing the processed data set into a training set and a testing set, training the XGBoost model by means of the training set, and finally obtaining the protein interaction site by means of the XGBoost model. The protein interaction site identification method aims to overcome a defect of relatively high result analysis difficulty caused by different degrees of false positive and false negative characteristics in predicting the protein interaction site. The protein interaction site identification method can overcome defects above and furthermore can improve identification precision of the protein interaction site.

Description

technical field [0001] The invention relates to the technical field of bioinformatics analysis, and more specifically relates to a method for identifying protein interaction sites. Background technique [0002] In all cells, proteins are the most important building blocks, and the interaction between proteins is the most fundamental activity in most cellular functions. The protein-protein interaction constitutes an important part of the cell biochemical reaction network. The protein-protein interaction network is the main way to realize the regulation of biological information and the key factor in determining the cell fate. The study of protein-protein interactions is the basis for understanding life activities and is one of the most important research fields in the post-gene era. With the implementation of the Human Genome Project, the data in the protein sequence database has increased significantly, and the structure and interaction between proteins cannot be determined...

AI Technical Summary

Problems solved by technology

[0007] The purpose of the present invention is to overcome the disadvantages in the prior art that when predicting protein interaction sites, there are different degrees of "false positive" and "false negative" characteristics, which make the result analysis more difficult, and provide a protein interaction The identification method of the interaction site can avoid having different degrees of "false positive" and "false negative" features, and greatly reduces the space and time overhead, and also improves the identification accuracy of the protein interaction site

Images