Identification of dynamic protein complexes based on fruit fly optimization algorithm

Abstract

Protein complexes play a significant role in understanding cellular life in postgenomic era. Yet up to now, the existing protein complex detection algorithms are mostly applied to static PPI networks and their performance is not very ideal for the deficiency of low efficiency and sensitive to noisy data. In this paper, a novel algorithm named Fruit fly Optimization Clustering Algorithm (FOCA), is proposed to identify dynamic protein complexes by combining Fruit fly Optimization Algorithm (FOA) and gene expression profiles. Particularly, we first find the always active proteins by the stable interactions of the dynamic PPI network and detect protein complex cores from those always active proteins. Then, FOA is used to merge of the rest proteins in every dynamic sub-network to their corresponding protein complex cores. The experimental results on DIP dataset demonstrate that FOCA is very effective in detecting protein complexes than the state-of-the-art complex detection techniques.