Closing the loop in stereoselective catalysis with data-driven approaches

ongoing

Data-driven methods promise to enable a highly structured approach for the development of asymmetric catalytic reactions, founded both on experimental and computational data. In this new but underdeveloped method, a small number of experiments is performed and that data is used to make a mathematical model to predict how new ligands will behave. Such a model is based on calculated or measured physical descriptors of the ligand, and correlations obtained between enantioselectivity and such descriptors also provide mechanistic insight. In an iterative (“looped”) approach, the model’s predictions are tested experimentally, and fed back to make an improved model, which is again tested experimentally, until satisfactory stereoselectivity and yield is obtained. Importantly, the integration and feedback of computational and experimental data during the research process is a significantly more efficient approach to developing asymmetric catalytic reactions and also provides mechanistic insight as the reaction is developed. The main research aim of this doctoral network is to develop powerful and readily applicable workflows for data-driven development of stereoselective catalysis. Using this data-driven approach requires a fundamentally different experimental workflow to developing catalytic reactions than is currently employed in most research laboratories. Since this requires a fundamental change in the way most experimental groups work, the data-driven approach is not widespread and remains underdeveloped. The next generation of chemists requires training in combined and integrated computational and experimental approaches, both in academia and industry. The main training aim of this doctoral network is to train researchers in comprehensive data-driven experimental approach for realizing challenging asymmetric catalytic methods.