Application of K-nearest neighbors in protein-protein interaction prediction

Abstract

Protein-protein interactions (PPIs) are an important part of many life processes in organisms. Almost all life processes are related to protein-protein interactions, and the study of protein interactions plays an important role in revealing the mysteries of life activities. In order to improve the prediction performance of protein-protein interaction, we are based on K-Nearest Neighbor (KNN), combined with protein sequence coding methods such as Conjoint Triad (CT), Auto Covariance (AC) and Local Descriptor (LD) to construct KNN-CT, KNN-AC and KNN-LD three prediction models of PPIs. The results show that the prediction models KNN-CT and KNN-AC have obtained accuracy rates of 94.29% and 94.69%, respectively, which are better than existing methods. The results show that K-nearest neighbors can be a useful complement to protein-protein interactions.