PPCM: Combing Multiple Classifiers to Improve Protein-Protein Interaction Prediction

Figure 1

The PPCM pipeline for protein-protein interaction prediction. Given a pair of query proteins QA and QB, their interaction possibility was first predicted by each of the 194 classifiers from GO2PPI and Phyloprof. Then, the classification scores were merged using Random Forests algorithm to generate the final PPI prediction score. Nine PPI classification scores were provided by PPCM. “SC” represents PPI networks in Saccharomyces cerevisiae. “Cross” represents all PPI networks except SC. “All” represents all PPI networks in both SC and cross species.