Modularized Random Walk with Restart for Candidate Disease Genes Prioritization∗

Identifying disease genes is very important not only for better understanding of gene function and biological process but also for human medical improvement. Many previous methods are based on modular nature of human genetic disease and the similarity between known disease genes and candidate genes. In this paper, we propose the method of Modularized Random Walk with Restart (MRWR) based on the functional module partition and module importance. Genes are prioritized in each functional module and then the gene ranking in each module is fused into a global ranking in the entire network. MRWR is applied to prostate cancer network. It is surprising that twenty-eight out of top fifty ranking genes are confirmed by PDGB or KEGG or literatures. MRWR significantly improves the performance of previous classical algorithms.