Predicting protein function based on the topological structure of protein interaction networks

摘要

With the completion of genome sequencing of several model organisms, the major challenge in the post-genomic era is to determine protein function at the proteomic scale. There are various approaches available for deducing the function of proteins of unknown function by protein information. In this paper, a new computational method is described, which provides the functional clustering of proteins on the basis of protein-protein interaction data. The hidden topological structures of complicated protein-protein interaction networks are revealed from graph theory. To validate the method, the yeast Saccharomyces cerevisiae protein-protein interaction network is analyzed. Compared with the current protein function prediction based on these networks, our method can substantially improve the quality of prediction with multiple data sources. The result shows that the proposed algorithm can successfully predict the correct functions in the less detailed classification, and the accuracy rate is about 92%.