Abstract: Identification of protein-protein interaction sites is essential for the mutant design and prediction of protein-protein networks. This paper proposes a method for predicting protein-protein interaction sites by combining support vector machine (SVM) and the sequence profiles, the accessible surface area (ASA) and the evolution rate of a residue. The dataset is trained and tested using 10-fold cross-validation. Accuracy of the proposed method is 72.91%, 5.71% higher than that of the method only using the sequence profiles and the evolution rate of a residue.

Key words: protein-protein interaction sites, support vector machine (SVM), sequence profiles, accessible surface area, evolution rate