A survey of computational method in protein-protein interaction research

With molecular biology research coming into the post-genome era focusing on proteomics, protein-protein interaction (PPI) has become an important topic of proteomics. The computational methods are widely used to analyze PPI data and guide experimental design for biologists, because they only need lower cost and have a shorter experimental period. In two aspects of constructing PPI network and analyzing it, all kinds of computational methods are surveyed. When it comes to constructing PPI network, some technologies, which use machine learning methods, are introduced, such as predicting PPI from protein sequence, expression or PPI network, mining PPI from biological medical literature database, evaluating PPI from various PPI data sets and so on. For PPI network analysis, three important and typical network parameters, (network diameter, degree distribution and cluster coefficient of node), and four models are illustrated in detail. Two PPI network modules with biological significance are graph clustering and network motif, and corresponding algorithms which use graph theory methods, are deeply analyzed. The concept and method about network alignment is also introduced. Finally, the computational research of protein-protein interaction is summarized and prospected.