Breaking Ties in Weighted Interactomes

Abstract

Automatic signaling pathway reconstruction methods using protein-protein interactomes are hindered by the unit/coarsely-weighted edges that lead to many paths sharing the same cost. We propose to use gene expression as an orthologous dataset from the pathway of interest to re-prioritize tied paths. The proposed method promotes/demotes paths based on the number of inferred true pathway edges. In general, it can be applied to any signaling pathway reconstruction method that outputs an ordered list of paths, trees, or other subgraphs.