Modularity Function And Identifying Algorithm Of Protein Complexes

Protein complex is very important for studying cell activity. Therefore, it is animportant work to promote the development of bioinformatics, efficientlyidentifying protein complexes from the protein-protein interaction networks (PPI).The topology of PPI networks is very complex, which contains many closelyconnected modules, these modules also have special biological significance. Thus,mining tap tight connected modules in the network is an important way to solve theproblem of identifying protein complexes. In this paper, the core problem is toidentify protein complexes, the main contents are as follows:(1) To biological characteristics of protein complex, we analyze the knownprotein complex topology in this paper. Concluded that: there are a lot ofoverlapping, small protein complexes in the PPI networks.(2) In order to identify modules from the network, Newman and someone elseproposed modularity function Q to measure module division quality of the network.However, after analysis we realize function Q has some shortcomings. Therefore, inthis article, we propose modularity function PQ for the identification of proteincomplexes, which is based function Q.(3) Since PPI networks are large-scale and complex, so the PPI networks initialmodule selection method is proposed to improve the efficiency of identifyingprotein complexes. By the way, BMM algorithm is proposed based on function PQ.In experiment, BMM algorithm is used to run real yeast PPI network data, andmatched the known protein complexes and function modules with identified proteincomplexes, the results show BMM is better than other algorithms.(4) Multifunctional proteins have important biological significance, usuallythey exist in several protein complexes or functional modules. In this paper, wematch identified protein complexes with functional annotation protein data, andverify a lot of multi-functional proteins, and identify a number of proteins withoutannotation.