Making better decisions during synthetic route design: leveraging prediction to achieve greenness-by-design

There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.

Abstract

A conceptual framework for incorporating machine learned ligand prediction into predictive route comparisons, to enable greener chemistry outcomes.

Abstract

Modern pharmaceuticals are becoming increasingly complex. Incorporating knowledge of a route's holistic sustainability during the route design process could be a critical enabler to minimizing the environmental impact of pharmaceutical manufacturing. The pursuit of the optimal synthesis has historically been characterized by disconnection strategy, or things like step count, however, the optimal synthesis of a molecule may also be assessed through environmentally relevant metrics. The synthesis with the lowest possible cumulative process mass intensity (cPMI) could be considered optimal, a route which may not necessarily be the shortest, but has the best holistic sustainability (for example, considering the synthesis of all reagents and reactants). Previously, we demonstrated the importance of assessing the entire synthetic network by including “above-the-arrow” reagents/reactants into cPMI, to reflect the impact of reagents, such as ligands, on the overall sustainability of the route. Here we present the development of a machine learning approach, using substrate fingerprints, to build a multiclass predictive model to identify which ligands will likely function in a Pd-catalyzed C–N coupling reaction. The resulting predicted multiclass probabilities were then linked to the corresponding ligand cPMIs to yield a probability-weighted predicted holistic PMI for the transformation, integrating the synthesis of the ligand. This proof-of-confidence study may extend our ability to holistically assess different synthetic route options, considering their full impact, to aid decision-making during route ideation. This may lead to greener outcomes in the development of synthetic routes in the pharmaceutical sector and beyond.

Related collections

Most cited references 35

Record: found
Abstract: found
Article: not found

Planning chemical syntheses with deep neural networks and symbolic AI

Mike Preuss, Mark Waller, Marwin H. S. Segler (2018)

To plan the syntheses of small organic molecules, chemists use retrosynthesis, a problem-solving technique in which target molecules are recursively transformed into increasingly simpler precursors. Computer-aided retrosynthesis would be a valuable tool but at present it is slow and provides results of unsatisfactory quality. Here we use Monte Carlo tree search and symbolic artificial intelligence (AI) to discover retrosynthetic routes. We combined Monte Carlo tree search with an expansion policy network that guides the search, and a filter network to pre-select the most promising retrosynthetic steps. These deep neural networks were trained on essentially all reactions ever published in organic chemistry. Our system solves for almost twice as many molecules, thirty times faster than the traditional computer-aided search method, which is based on extracted rules and hand-designed heuristics. In a double-blind AB test, chemists on average considered our computer-generated routes to be equivalent to reported literature routes.