Graph Partitioning Method for Functional Module Detections of Protein Interaction Network

Study on topology structure of protein interaction network has been suggested as a potential effort to discover biological functions and cellular mechanisms at systems level. In this work, we introduced a graph partitioning method to partition protein interaction network into several clusters of interacting proteins that share similar functions called functional modules. Our proposed method encompasses three major steps which are preprocessing, informative proteins selection and graph partitioning algorithm. We utilized the protein-protein interaction dataset from MIPS to test the proposed method. We use gene ontology information to validate the biological significance of the detected modules. We also downloaded protein complex information to evaluate the performance of our method. In our analysis, the method showed high accuracy performance indicates that this method capable to detect highly significance modules. Hence, this showed that functional modules detected by the proposed method are biologically significant which can be used to predict uncharacterized proteins and infer new complexes.