Detection of protein-protein interactions from amino acid sequences using a rotation forest model with a novel PR-LPQ descriptor

Protein-protein interactions (PPIs) play an essential role in almost all cellular processes. In this article, a sequence-based method is proposed to detect PPIs by combining Rotation Forest (RF) model with a novel feature representation. In the procedure of the feature representation, we first adopt the Physicochemical Property Response Matrix (PR) method to transform the amino acids sequence into a matrix and then employ the Local Phase Quantization (LPQ)-based texture descriptor to extract the local phrase information in the matrix. When performed on the PPIs dataset of Saccharomyces cerevisiae, the proposed method achieves the high prediction accuracy of 93.92 % with 91.10 % sensitivity at 96.45 % precision. Compared with the existing sequence-based method, the results of the proposed method demonstrate that it is a meaningful tool for future proteomics research.