Firefly clustering method for mining protein complexes

Abstract

It is a hot research to explore protein complexes which are closely related to biological processes from the biological network. As a novel swarm intelligence optimization algorithm, the firefly algorithm (FA) has been verified to solve many optimization problems. In this study, we transform the protein clustering problem into an optimization problem in protein-protein interaction (PPI) network. A new method for mining protein complexes based on the firefly algorithm was proposed, called FC. A new objective function was proposed to find the high cohesion and low coupling clusters. A thorough comparison completed for different protein clustering methods has been carried out. The clustering results show that FC method outperforms the other state-of-the-art methods in accuracy of detecting complexes from PPI network.