S_NAr Regioselectivity Predictions: Machine Learning Triggering DFT Reaction Modeling through Statistical Threshold

Abstract

Fast and accurate prospective predictions of regioselectivity can significantly reduce the time and resources spent on unproductive transformations in the pharmaceutical industry. Density functional theory (DFT) reaction modeling through transition state theory (TST) and machine learning (ML) methods has been widely used to predict reaction outcomes such as selectivity. However, TST reaction modeling and ML methods are either timeconsuming or data-dependent. Herein, we introduce a prototype seamlessly bridging ML and TST modeling by triggering resource-intensive but much less domain-sensitive DFT calculations only on less confident ML predictions. The proposed workflow was trained and tested on both the Pfizer internal dataset and the USPTO public dataset to predict regioselectivity for S N Ar reactions. Our method is accurate and fast, which achieves 96.3 and 94.7% accuracy in predicting the correct major product on Pfizer and USPTO datasets, respectively, in a fraction of conventional TST computing time.

Dataset Design for Building Models of Chemical Reactivity

“…most often corresponds to the difference between enantio-determining transition states with the general reaction mechanism otherwise being the same, allowing one to neglect factors that confound modeling yield as an output, such as side reactions or differences in turnover rates of a catalyst. Likewise, regioselectivity is an internally consistent metric that relies only on direct comparisons between candidate atom sites. − …”

Carbon–Nitrogen Bond Formation Using Sodium Hexamethyldisilazide: Solvent-Dependent Reactivities and Mechanisms

“…S N Ar reactions (entries 15–19) represent a reaction class of unquestioned importance . We find it somewhat confounding that the S N Ar reactions seem to be more effective with soft nucleophiles.…”

“…Aminolyses illustrated with generic examples in Scheme are legion. S N Ar substitutions and peptide bond formations are of unquestioned importance in pharmaceutical chemistry. , With that said, any reaction in which the electrophile is merely heated in ammonia or simple alkylamine will be difficult to improve upon. However, should these simple protocols fail, owing to low reactivity or poor selectivity, the experimentalist is left with few options.…”

Regio-MPNN: Predicting Regioselectivity for General Metal-Catalyzed Cross-Coupling Reactions using Chemical Knowledge Informed Message Passing Neural Network

“…After that, they focused on the regioselectivity problem in nucleophilic aromatic substitution (SNAr) reactions and attached a checker at the end of their model to distinguish the possible products with similar predictive scores. 24 In Ree's, Tomberg's, and Guan's work, they all treated the electronic properties as the major factors for regioselectivity in their interested reactions and did not take steric effects into account explicitly. Currently, there is no predictive model available that combines both electronic and steric effects with data-driven deep learning methods.…”