CYTOSCAPE RETRIEVED PROTEIN-PROTEIN INTERACTION (PPI) NETWORKS: SEEKING THE CORRELATION OF HUMAN PROTEINS' TOPOLOGICAL FEATURES BETWEEN MAJOR PUBLIC PPI DATABASES DUE TO THEIR MEDICAL IMPORTANCE

Background: Protein-protein interaction (PPI) databases have become major sources to study networks and cellular pathways, which can be utilized to get invaluable information in biomedical researches. There are some public PPI databases have already deposited the large amount of experimentally verified interactions from literatures using different curation policies. Our aim was to evaluate to what extent common human proteins' topological features have correlations between these databases. Methods: Five major public PPI databases (DIP, MINT, HPRD, IntAct and BIND) obtained from Cytoscap. Using statistical analysis such as calculation of Correlation coefficients, seven topological features of some common human proteins were compared between these databases. Results: The results showed that some medium and weak significant correlations of shared proteins' topological features were between five databases, and one robust correlation for degree was found between HPRD and BIND (Spearman's rank r = 0.79, P = 0) databases. Conclusions: Several lines of evidence indicated that PPI data have been reported according to different organizing strategies of databases; as a result, this issue may affect the structure of derived PPI networks. Probably because of this reason, our results confirm this point that topological features of proteins in current human PPI networks are highly network dependent.