A Graph-based Method For Alignment Of Protein Interaction Networks

With advances in the development of bioinformatics, our research focuses on thebiological macromolecules such as protein and DNA. Proteins play an important role inthe organisms, and protein-protein interactions are the basis of many processes thatmaintain cellar structure and function. So it is of great significance to do research inprotein interaction networks. An important research is alignment analysis ofprotein-protein interaction (PPI) networks. We can predict protein functions and obtainconserved modules in this way.A key problem in alignment of PPI networks is how to define protein similarities.Usually it is computed by protein sequences alignment, but in this paper we describe anew method, GSim, based on graphs to compute protein similarity. It combinessequences similarities and the property of PPI networks. We find that it works better inthe experiments.There is a network alignment algorithm that converts alignment problem into anEigen value problem of a matrix. This algorithm solves the vector of protein similarities,then exerts protein mapping form the vector, finally searches conserved modules. Thispaper improves this algorithm in adding protein sequence similarities. We doexperiments between D. melanogaster and S. cerevisiae PPI networks in two situations.