Inference of protein function from the structure of interaction networks

Abstract

We consider the problem of using graph-theoretical techniques to predict the function of unannotated proteins in an organism's proteome. Specifically, we present an overview of the major methods for predicting protein function based on interaction network structure and describe an abstract framework within which these methods can be treated in a unified fashion. We also present a comparison of the proposed methods and highlight some open theoretical and practical questions in the area. © 2011 Springer Science+Business Media, LLC.