Large Scale Ranking and Repositioning of Drugs with Respect to DrugBank Therapeutic Categories

The ranking and prediction of novel therapeutic categories for existing drugs (drug repositioning) is a challenging computational problem involving the analysis of complex chemical and biological networks. In this context we propose a novel semi-supervised learning problem: ranking drugs in integrated bio-chemical networks according to specific DrugBank therapeutic categories. To deal with this challenging problem, we designed a general framework based on bipartite network projections by which homogeneous pharmacological networks can be combined and integrated from heterogeneous and complementary sources of chemical, biomolecular and clinical information. Moreover, we propose a novel method based on kernelized score functions for fast and effective drug ranking in the integrated pharmacological space. Results with 51 therapeutic DrugBank categories involving about 1300 FDA approved drugs show the effectiveness of the proposed approach.