Global Topological Study of the Protein-protein Interaction Networks

We employed the random graph theory approach to analyze the protein-protein interaction database DIP (Feb. 2004), for seven species (S. cerevisiae, H. pylori, E. coli, C. elegans, H. sapiens, M. musculus and D. melanogaster). Several global topological parameters (such as node connectivity, average diameter, node connectivity correlation) were used to characterize these protein-protein interaction networks (PINs). The logarithm of the connectivity distribution vs. the logarithm of connectivity indicated that PINs follow a power law (P(k) ~ k) behavior. Using the regression analysis method we determined that γ lies between 1.5 and 2.4, for the seven species. Correlation analysis provides good evidence supporting the fact that the seven PINs form a scale-free network. The average diameters of the networks and their randomized version are found to have large difference. We also demonstrated that the interaction networks are quite robust when subject to random perturbation. Average node connectivity correlation study supports the earlier results that nodes of low connectivity are correlated, whereas nodes of high connectivity are not directly linked. These results provided some evidence suggesting such correlation relations might be a general feature of the PINs across different species.