Artificial Intelligence-directed Reaction Discovery

Artificial intelligence (AI) and computational modelling have much potential to inform the development of synthetic chemistry. Applications range from improving catalysts to enable sustainable manufacturing, to the discovery of new and sought after chemical reactivity. We will focus on the latter, specifically the development of methods to prepare complex molecules using organoboron building blocks. Ultimately, this will allow new areas of chemical space to be explored for applications like pharmaceutical and agrochemical discovery, and will be of interest to chemists in both academia and industry.

We will use a synergistic approach of organic synthesis, computational chemistry and AI to expedite the development of new methods. The AI-driven prediction of reaction pathways will be based on proposing and sampling possible mechanisms using graph theoretical methods and semi-empirical quantum chemistry, as a precursor to higher level modelling and synthesis. We will feedback data from experiment and theory into each process, enabling continuous refinement of both the model and synthetic method.

This approach will be applied to the design of new palladium-catalysed reactions of alkylboronic esters. The field of transition metal catalysis is ideal to take advantage of this synergistic approach because well defined parameters, such as catalyst structure, can be suitably modelled.