Development Of Novel Computational Methods For Drug Discovery And Repurposing In Oncologic Diseases And Other Illnesses

Targeting disease-related proteins is important for drug discovery, yet target-based method have not been fruitful. Bottlenecks involve: (1) establishing biologically valid drug- protein target associations, and (2) assessing the physiologic effects of those interactions at the systems level. Here we develop novel computational methods for overcoming these challenges. For the accurate prediction of drug-target interactions, we investigate the ability of two independent proteochemometric methods entitled R-TMFS and ES-Screen to prioritize known drug binders over decoys for diverse sets of protein targets. R-TMFS is a docking-based method that incorporates molecule shape and physicochemical properties, whereas ES-Screen is an electrostatics-driven method that accentuates the role of electrostatics in biomolecular recognition and binding kinetics. R-TMFS and ES-Screen are also used to predict previously un-reported kinase targets for the anti-hookworm medication mebendazole. Follow-up in vitro binding assays confirm mebendazole inhibition of multiple kinases such as ABL1, JAK2, JNK3, and RAF1, attesting to the repurposing potential of mebendazole for various cancers. For higher-order physiologic contextualization of drug-target signatures, we devised a computational platform named NET-TMFS that annotates drugs with biological endpoint effects including protein-protein interactions, signaling pathways, molecular functions, and disease effects. NET-TMFS recapitulated over 50 drug-disease, 100 drug-pathway, and drug-PPI associations established in the literature. NET-TMFS also predicted potential carcinogenic effects of the cholesterol-lowering drug ezetimibe, a phenomenon documented in clinical trials. In summary, we have developed novel computational methods for addressing major bottlenecks in the drug discovery process. We hope our methods will aid in finding effective therapeutics for many diseases with greater efficiency and lower costs.