Predicting Three-Component Reaction Outcomes from 40k Miniaturized Reactant Combinations

Efficient drug discovery relies on accessing diverse small molecules expediently and reliably. Improvements to reliability through machine learning predictions are hampered by poor availability of high-quality reaction data. Here, we introduce an on-demand synthesis platform based on a three-component reaction that delivers drug-like molecules overnight. Miniaturization and automation enable the execution and analysis of 50,000 reactions on a 3 microliter scale with distinct substrates, producing the largest public reaction outcome dataset. With machine learning, we accurately predict the result of unknown reactions and analyze the impact of data set size on model training. This study advances the on-demand synthesis of drug-like molecules through concatenating chemoselective reactions and provides a sufficiently large data set to critically evaluate emerging machine learning approaches to predicting chemical reactivity.