Predicting yeast synthetic lethal genetic interactions using protein domains

Synthetic lethal genetic interaction (SLGI) is an important biological phenomenon. Such interactions are of interest as they can be used to predict function of unknown proteins and find drug targets or drug combinations. High throughput biological experiments enhance the capability in identifying genetic interactions, but the large amount of protein pairs still make the task of genome-wide identification of genetic interactions overwhelming. Computational based prediction of SLGIs is promising but still hampered by the unclear molecular mechanism of SLGIs.;Protein domains with conserved functions serve as the building blocks of proteins. The genetic interaction that occurs between a pair of proteins could be essentially related to or even dominated by the domains underneath. We applied support vector machine (SVM) classifier and maximum likelihood estimation (MLE) method to predict SLGIs in yeast based on domain information in proteins. Our study demonstrates that yeast SLGIs could be explained by the genetic interactions between domains of those proteins. Moreover, we retrieved a set of polypeptide clusters and used them for the prediction of SLGIs. Besides providing better performance, this approach allows us to predict genetic interactions in a more general fashion.;We proposed a novel idea for the prediction of SLGIs, upon which multiple approaches were derived. This study helps the understanding of originality of functional relationship in SLGIs at the domain level that may significantly aid the biology community in further analysis of genetic interaction related studies.