Protein Complexes Identification Based On Communication Theory

In the post-genomic era, the focus of biomedical research has transited from traditional medical genomics to proteomics. One of the most important challenges is to systematically analyze how the proteins accomplish the life activities by interacting with each other. In particular, Identification of protein complexes is important to reveal specific biological processes and predict protein function.Based on analyzes the structure of the protein network function module, according to communication theory, some excellent methods are proposed for identifying protein complexes, the main original works include:With the expansion of the scale of protein-protein interaction data, local information network based clustering algorithm is increasingly being adopted by researchers. This kind of methods do not concern the problem from a global perspective, it uses only the local information of nodes. Besides, this kind of methods usually can be done effectively in a relative short period of time and it is more suitable for the growing protein network. To solve the above problems protein complexes recognition GO-TSA based on communication theory is presented, each node in the network is considered as a source, each node spread signals to each other by connecting relations, and every iteration of the spread process, Node will be updated according to the strongest signal type of its neighbors. Experimental results show that the algorithm can clustering the protein networks in a very short period of time. Compared with the algorithm CPM (k=3) and algorithm Core-Attachment, In almost every case, our method has a better performance.For the case that a protein may belong to multiple protein complexes, on the basis of the algorithm GO-TSA, we presents a complex overlapping mining algorithms GO-OTSA, it introduces a new signal structures (c, b) in the algorithm, Whether a signal typed is kept depends on the balance of the node signal and max signal ratio. Experimental results show that the performance of the algorithm GO-OTSA has improved.