Protein complex detection from PPI networks on Apache Spark

![CDATA[Protein-Protein Interaction (PPI) network is a network of biomolecular interactions which plays a major role in modeling and analyzing biological activities. Studies of functional modules from PPI networks provide a better understanding of biological mechanisms. Recent advances in both biological and computer sciences demands for the vast amount of PPI networks data to be processed by experimental and computational methods. This could be a great challenge to find functional modules within these large networks. Existing methods are used to identify the functional modules, but some of them do not consider overlapping between functional module clusters. Moreover, most of the methods run on a single machine. Also, many existing algorithms only focus on topological features of PPI networks. In this paper, we introduce a new way for detecting the functional modules. It considers overlapping between clusters and runs on Apache Spark - a distributed processing platform. Our algorithm also considers both topological and biological features of PPI networks. The evaluation results show improved execution speed as well as more accurate results compared to classic methods.]]