Drug Repurposing is one of the most promising new avenues in drug discovery, which offers treatment potential for various diseases including cancer treatment. This research explores drug repurposing with computational deep learning techniques for Metformin and Aspirin, two most commonly well-known drugs, for their potential efficacy with cancer treatments. In addition, Cisplatin, a current chemo drug was also studied to compare results. DeepPurpose, a Deep Learning based open-source toolkit that outputs protein binding scores was used. For this, Metformin, a type-2 diabetes drug, used all over the world, to be repurposed for cancer, Aspirin, a common non-steroidal anti-inflammatory drug to be repurposed for cancer, and Cisplatin, a chemo drug were studied for their binding scores, a measure of their suitability as anticancer agents. We used three different models on two benchmark datasets, namely KIBA and DAVIS. The research was carried after developing custom python scripts run on a Jupyter Notebook. Cross-validation of binding affinity results was compared with a BindingDBIC50 dataset. MSE and Confidence Indices were also computed. The relationship between the drug targets was also studied using a string diagram to identify interactions between them as well as any associated cancer pathways. After running the simulations, it was observed that ‘KRAS’ and ‘HER2’ targets have the highest binding scores for Metformin, and Cisplatin had the highest binding scores of all the three drugs. String interactions were also studied to identify relevant pathways. Metformin and Aspirin show the potential to be as inhibitors for certain specific cancer targets, as a first step in this kind of research and if used in combination with other cancer drugs, they could be more effective cytotoxicity agents for cancer treatment. Further research is needed to take it to the next steps.
Vokkarne et al. (Sat,) studied this question.