Abstract:Protein-protein interactions (PPIs) networks play an important role in most of biological processes. Although much effort has been devoted to using high-throughput biological technologies to identify PPIs of various kinds of organisms, the experimental methods are expensive, time-consuming, and tedious. Therefore, developing computational methods for predicting PPIs is of great significance in this post-genomic era. In recent years, the exponential increase of available protein sequence data leads to the urgent need for sequence-based prediction model. In this paper, we report a highly efficient method for constructing PPIs networks. The main improvements come from a novel protein sequence representation called pseudo-SMR, and from adopting weighted sparse representation based classifier (WSRC). When predicting the PPIs of Yeast, Human and H. pylori datasets, the 5-fold cross-validation accuracies performed by the proposed method achieve as high as 97.09%, 96.71% and 91.15% respectively, significantly better than previous methods. To further evaluate the performance of the proposed method, extensive experiments are performed to compare the proposed method with state-of-the-art Support Vector Machine (SVM) classifier. Promising results obtained show that the proposed method is feasible, robust and powerful.

Construction of Reliable Protein-Protein Interaction Networks Using Weighted Sparse Representation Based Classifier with Pseudo Substitution Matrix Representation Features