home · mobile · calendar · colloquia · 2004-2005 · 

Colloquium - Ben-Hur

A Kernel Method for Predicting Protein-Protein Interactions
University of Washington

Most proteins perform their function by interacting with other proteins. Therefore, information about the network of interactions that occur in a cell can greatly increase our understanding of protein function. We present a kernel method for predicting protein-protein interactions using a combination of data sources, including protein sequences, Gene Ontology annotations, local properties of the network, and interactions in different species. We propose a pairwise kernel that provides a similarity between pairs of proteins, and illustrate its effectiveness in conjunction with a support vector machine classifier. We obtain improved performance by combining several sequence-based kernels based on k-mer frequency, motif and domain composition and by further augmenting the pairwise sequence kernel with features that are based on additional sources of data. In yeast, at a false positive rate of 1%, the classifier retrieves close to 80% of a set of trusted interactions, demonstrating the ability of our method to make accurate predictions despite the sizable fraction of false positives that are known to exist in interaction databases.

Department of Computer Science
University of Colorado Boulder
Boulder, CO 80309-0430 USA
May 5, 2012 (14:13)