Seed-weighted random walks ranking method and its application to leukemia cancer biomarker prioritizations

The identification of effective biomarkers for preventive intervention or targeted therapies will increase survival rate of cancer patients dramatically. However, the unclear molecular mechanism of carcinogenesis still blocks the discovery process of effective cancer biomarkers. Network-based analyses have been introduced into computational biomarker discovery for many years. The random walks ranking (RWR) algorithm is one of the most successful methods, which use known cancer susceptivity genes as seeds, and exploit global network topology to prioritize proteins in a protein-protein interaction (PPI) network. In this paper, we proposed a modified method - seed weighted RWR (SW-RWR) for prioritizing cancer biomarker candidates, which used the information of cancer phenotype association for assigning each seed gene a weighted value to guide RWR algorithm in a global human PPI network. From a case study in leukemia, SW-RWR outperformed a local network analysis in coverage and also showed better accuracy and sensitivity than the original RWR method. The result suggests that the tight correlation among different cancer phenotypes could play an important role in cancer biomarker discovery.