Novel domain identification approach for protein-protein interaction prediction

Protein-protein interaction (PPI) plays a crucial role in cellular biological processes and functions. The identification of protein interactions can lead to a better understanding of infection mechanisms and the development of many drugs and treatment. Several physiochemical experimental techniques have been applied to identify PPIs. However, these techniques are significantly time consuming and have covered only a small portion of the complete PPI networks. As a result, the need for computational techniques has increased to validate experimental results and to predict novel PPIs. In this work, we propose a domain identification approach for PPI prediction based only on Amino Acid (AA) sequence information. Domains are the structural and functional units of proteins and, therefore, proteins interact as a result of their interacting domains. We identify structural domains within proteins through integrating the amino acid compositional index in conjunction with physiochemical properties to construct a domain linker profile which is used to train a TreeBagger domain identification predictor. Once structural domains are identified in two protein sequences, we predict whether these two proteins interact or not by analyzing the interacting structural domains that they contain. The proposed approach was tested on a standard PPI dataset and showed considerable improvement over the existing PPI predictors.