Evolutionary Graph-Clustering vs Evolutionary Cluster-Detection Approaches for Community Identification in PPI Networks

Community detection in protein-protein interaction networks (PPIs) is an active area of research, and many studies have applied Genetic Algorithms (GAs) to this problem. This paper summarizes the different GA based approaches for community detection in PPIs and provides a taxonomy of these methods. Detailed comparative studies are then provided comparing an evolutionary graph-clustering approach (EGCPI) based on the partitioning paradigm and an evolutionary cluster-detection approach based on an evolutive and incremental search for potential communities in the graphs (GA-PPI-Net). The communities obtained by the two algorithms on Collins PPI network are compared according to the average similarity and interaction between genes, and also according to the recovery rate of known communities in some biological pathways. Experiments tests verify the effectiveness of the GA-PPI-Net approach compared with EGCPI approach.