Getting poster data...
Kubra Karagoz, Kazım Yalçın Arga (Department of Bioengineering, Marmara University, Goztepe 34722, Istanbul, Turkey )Protein-protein interactions play a key role in many cellular processes. Recently, 140385 physical protein-protein interactions between 6344 proteins are presented in literature via several non-commercial databases. However, the experimental data obtained mostly from high-throughput experiments is liable to contain large number of false positives. The main objective of this study is to represent an almost complete, highly reliable protein-protein interaction network of Saccharomyces cerevisiae via identification of true positives using a Bayesian approach by integrating various genomic features from different biological level, i.e. database information, type of experimentation, PubMed identifiers, gene ontology (GO) annotations, structural domain information, and gene expression correlation. Our analysis resulted with a broad score distribution and a highly-reliable network consisting of 19703 interactions with scores higher than the threshold score of 0.90, for which the true positive and false positive rates were 86% and 21%, respectively.