Data‐driven prediction of therapeutic efficacy and optimization of de novo drug combinations on the human protein‐protein interaction network (842.11)
Hann Wang · The FASEB Journal · 2014
Combination therapy is a promising direction for cancer research as it is shown to eradicate drug‐resistant cancer more effectively than monotherapy. The challenge, however, in creating a successful combinatorial treatment lies in the inability to screen through a huge number of possible combinations. We introduce PROPHECY (Predictive Optimization of Pharmaceutical Efficacy), a data‐driven search engine for the selection of drugs in combination therapy. The user provides a genetic profile, e.g. mutations in cancer cell lines or primary cells, and PROPHECY will select optimal drug combinations from a comprehensive library of existing drugs to meet clinical objectives. The decision is made by comparing the efficacy of drug combinations with in silico screening, in which a predictive machine is trained with a large‐scale database of drug screening experiments to recognize high‐order target‐gene interactions in the human protein‐protein interaction network (PPI). The predictive module in the machine incorporates knowledge of information propagation in the network, so can accurately elucidate interactions of a group of drug targets and disease genes in a de novo drug combination. Unlike other approaches of reductionistic drug development, PROPHECY can fully acknowledge the synergy between multiple drugs and predict the most effective multi‐target treatment against drug resistant cancer. Grant Funding Source : Supported by NSF