Please join Yale Chemistry for a thesis seminar with Abbigayle Cuomo, Newhouse Group.
Title: Data-Driven Synthetic Planning: From Selectivity Prediction to Complexity-Aware Route Design
Summary: An integrated framework couples global, literature-scale learning with local, mechanistically delimited models to yield reliable, interpretable predictions for synthesis planning. Key contributions include a feed-forward neural network for the prediction of enantioselectivity in the Negishi reaction, SHAP-guided descriptor models for site selectivity in C–H oxidation, and an EvolvedComplexity metric for route evaluation, applied to the redesign of routes to important pharmaceutical APIs. Across benchmarks and prospective case studies, these approaches improve predictive accuracy, prioritizes informative experiments, and shorten and de-risk synthetic sequences, linking statistical prediction to mechanistic insight and executable process chemistry.
