An efficient method to identify essential proteins for different species by integrating protein subcellular localization information

Essential proteins are indispensable to maintain life activities in living organisms, and play important roles in the studies of pathology, synthetic biology, and drug design. Many computational methods are employed to identify essential proteins from Protein-protein Interaction Networks (PINs). In this paper, considering the different importance of protein-protein interactions which take place in different subcellular compartments, a Compartment Importance Centrality (CIC) method is proposed to detect essential proteins by integrating protein subcellular localization information. The experiments were carried on four species (Saccharomyces cerevisiae, Homo sapiens, Mus musculus and Drosophila melanogaster), and the performance of CIC was compared with other centrality methods, including the centrality methods solely based on topology and the ones combining both topology and other biological knowledge. The results show that CIC method has better performance to predict essential protein on four species. Furthermore, different from methods which overfits with the features of essential proteins of one species and may perform poor for other species, CIC has a wide applicable scope to identify essential proteins for different species.