Prediction of Protein-protein Interaction Sites at Interface Topology Level

Protein-protein interactions play a crucial role in many biological processes such as immune response, enzyme catalysis, and signal transduction. Identifying the interacting sites between the proteins is important for understanding the functional mechanisms, and is crucial for drug development. Interaction profile hidden Markov model (ipHMM) was shown to be an effective tool for modeling protein-ligand interaction site in previous study. In this study, the ipHMM was applied to predict protein-protein interaction site by taking into account of interacting partner and topology information. Particularly, it was found that the performance of ipHMM at domaindomain interaction (DDI) family level was significantly lower for DDI families with multiple topology interfaces. To address this problem, we proposed to develop ipHMM at DDI interface topology level. The performance of interacting site prediction was significantly improved. The average sensitivity/recall was improved from 36.6% to 74.0%. The average precision was improved from 49.1% to 75.9%. The Matthews correlation coefficient was improved from 46.4% to 77.3%.