Abstract

To predict protein-protein docking sites in a massive protein dataset, we built a cloud computing based computing pipeline. This pipeline conforms to Elastic MapReduce. The implementation of this pipeline includes three components. First, the cloud computing is based on the application of the open source hadoop platform. Second, the pipeline combines several existing protein-protein docking site methods. Third, the pipeline takes advantage of network computing resource to predict protein-protein docking sites by distributed data processing services. The results show our method can highly improve the performance of protein-protein docking site prediction.