A MapReduce-Based Parallel Random Forest Approach for Predicting Large-Scale Protein-Protein Interactions

The protein-protein interactions (PPIs) play an important part in understanding cellular mechanisms. Recently, a number of computational approaches for predicting PPIs have been proposed. However, most of the existing methods are only suitable for relatively small-scale PPIs prediction. In this study, we propose a MapReduce-based parallel Random Forest model for predicting large-scale PPIs using only proteins sequence information. More specifically, the Moran autocorrelation descriptor is firstly used to extract the local features from protein sequence. Then, the MapReduce-based parallel Random Forest model is utilized to perform PPIs prediction. In the experiment, the proposed method greatly reduces the required time to train the model, while maintaining the high accuracy in the prediction of potential PPIs. The promising results demonstrate that our method can be used as an efficient tool in the field of large-scale PPIs prediction, which greatly reduces the required training time and has high prediction accuracy.