Discovering Protein Complexes from Protein-Protein Interaction Data by Local Cluster Detecting Algorithm

Proteins are usually grouped into complexes to perform functional tasks in organisms. So discovery of protein complexes is critical to understand principles of cellular organization and function. Recently high-throughput techniques have produced large amount of protein-protein interaction (PPI) data, which make it possible to detect protein complexes automatically from PPI data by using computational method. In this paper, we propose a local cluster detection algorithm (LCDA) to find the protein complexes. Super to other methods existed, LCDA is able to find both non-overlapping and overlapping clusters. The experiment results show that LCDA can detect protein complexes with very high accuracies.