Global insights into protein complexes through integrated analysis of the interactome and knockout lethality

Harukazu Suzuki1, Rintaro Saito 2, Yoshihide Hayashizaki, RIKEN Genomic Sciences Center ; 2, RIKEN Genomic Sciences Center

In the last ISMB meeting, we reported the development of the interaction generality measure (IG1), a computational method to address an intrinsic problem that protein-protein interactions obtained from biological experiments often include many false positives. However, our IG1 method was a simple method for evaluating the reliability of interactions that did not consider the topological properties of the protein interaction network beyond the target pair of proteins. To overcome this inaccuracy, we developed IG2, in which principal-component analysis of the topological properties is incorporated into the evaluation of the reliability of the interaction. We evaluated the utility of the IG2 measure in several analyses. Next, we performed an integrated computational analysis of data derived from a comprehensive set of protein-protein interactions (interactome) and a phenotype dataset on lethality in Saccharomyces cerevisiae. For this analysis, we selected a reliable interactome data using IG2. Those efforts gave clear evidence that proteins with lethal phenotypes in knockout studies (lethal proteins) may interact with each other to form functional protein complexes to perform their cellular roles. In addition, we showed that interactions between lethal proteins are rather restricted to the same cellular pathway or function, and it is quite unlikely that they interact with other lethal proteins functioning in different cellular roles. Further, our results allowed us predictions on the functions of uncharacterized lethal proteins with an estimated 93% accuracy. Thus, the analysis described in here can provide global insights into the biological features of the protein complexes.