Prediction of protein-protein interactions from amino acid sequences using extreme learning machine combined with auto covariance descriptor

Abstract

Protein-protein interactions (PPIs) are crucial for almost all cellular processes, including metabolic cycles, DNA transcription and replication, and signaling cascades. Unfortunately, the experimental methods for identifying PPIs are both time-consuming and expensive. Therefore, it is important to develop computational approaches for predicting PPIs. In this paper, a sequence-based method was developed for identifying new protein-protein interactions (PPIs) by means of Extreme Learning Machine (ELM) combined with a novel representation using auto covariance (AC). The AC descriptors account for the interactions between residues a certain distance apart in the protein sequence, thus this method adequately takes the neighboring effect into account and enables us to extract more PPI information from the protein sequences. ELM is a kind of accurate and fast-learning innovative classification method based on the random generation of the input-to-hidden-units weights followed by the resolution of the linear equations to obtain the hidden-tooutput weights. When performed on the PPI data of Saccharomyces cerevisiae, the proposed method achieved 90.42% prediction accuracy with 90.12% sensitivity at the precision of 90.67%. Extensive experiments are performed to compare our method with state-of-the-art techniques Support Vector Machine (SVM). Achieved results show that the proposed approach is very promising for predicting PPI, and would make a helpful supplement to experimental approaches.