Identification of conserved protein complexes by module alignment

Recently, accumulating evidence suggests that biological systems are composed of interacting, separable, functional modules (e.g., protein complexes) - groups of vertices within which connections are dense while between which they are sparse. These functional modules always correspond to well-known protein complexes, which may be evolutionarily conserved across multiple species. Therefore, in this paper, we propose a method based on module alignment, which integrates protein interaction, and sequence information for finding conserved protein complexes. First, our method decomposes Protein-Protein Interaction (PPI) networks into modules by module detection methods, and then identifies conserved complexes by module alignment based on sequence similarity between pairs of proteins from each of the species. We test our method between Saccharomyces cerevisiae and Drosophila melanogaster. The results show that our method gets a higher accuracy for identification of conserved complexes.