Sarvesh Mehta received his MS Dual Degree in  Computational Natural Sciences (CNS). His research work was supervised by Prof. Deva Priyakumar. Here’s a summary of his research work on Machine Learning Framework for Efficient Chemical Space Exploration:

In drug discovery applications, high throughput virtual screening exercises are routinely performed to determine an initial set of candidate molecules referred to as “hits”. In such an experiment, each molecule from large small-molecule drug library are evaluated for physical properties such as the LogP, TPSA, and docking score against a target receptor. In real-life drug discovery experiments, the drug libraries are extremely large but still a minor representation of the essentially infinite chemical space , and evaluation of physical property for each molecule in the library is not computationally feasible. In the current study, a novel Machine learning framework for Enhanced MolEcular Screening (“MEMES”) based on Bayesian optimization is proposed for efficient sampling of chemical space. The proposed framework is demonstrated to identify 90\% of top-1000 molecules from a molecular library of size about 100 million, while calculating the docking score only for about 6\% of the complete library. We believe that such a framework would tremendously help to reduce the computational effort in not only drug-discovery but also areas that require such high-throughput experiments.