Prediction of protein-protein interactions

We used a naive Bayesian classifier to predict genome-wide protein-protein interactions in yeast by integrating information from different genomic features, ranging from co-expression relationships to similar phylogenetic profiles. At a certain level of sensitivity the predictions were more accurate than the existing high-throughput experimental dataset. We explored the limits of genomic data integration, assessing the degree to which the predictive power increases with the addition of more features. We are currently improving these results by applying recently developed machine learning algorithms to this problem.