Predicting Regioselectivity in Radical C−H Functionalization of Heterocycles through Machine Learning

• Published in: Angewandte Chemie International Edition Vol. 59; no. 32; pp. 13253 - 13259
• Main Authors: Li, Xin, Zhang, Shuo‐Qing, Xu, Li‐Cheng, Hong, Xin
• Format: Journal Article
• Language: English
• Published: Germany Wiley Subscription Services, Inc 03.08.2020
• Edition: International ed. in English
• Subjects: Accuracy; Computer applications; Heterocyclic compounds; Hydrogen bonds; Learning algorithms; Machine learning; mechanism-based computational statistics; Predictions; radical C−H functionalization; random forest model; Regioselectivity; regioselectivity prediction; Selectivity; machine learning; radical C−H functionalization; random forest model; mechanism-based computational statistics; regioselectivity prediction
• ISSN: 1433-7851; 1521-3773
• DOI: 10.1002/anie.202000959

Radical C−H bond functionalization provides a versatile approach for elaborating heterocyclic compounds. The synthetic design of this transformation relies heavily on the knowledge of regioselectivity, while a quantified and efficient regioselectivity prediction approach is still elusive. Herein, we report the feasibility of using a machine learning model to predict the transition state barrier from the computed properties of isolated reactants. This enables rapid and reliable regioselectivity prediction for radical C−H bond functionalization of heterocycles. The Random Forest model with physical organic features achieved 94.2 % site accuracy and 89.9 % selectivity accuracy in the out‐of‐sample test set. The prediction performance was further validated by comparing the machine learning results with additional substituents, heteroarene scaffolds and experimental observations. This work revealed that the combination of mechanism‐based computational statistics and machine learning model can serve as a useful strategy for selectivity prediction of organic transformations. Mechanism‐based computational statistics allowed the machine learning prediction of regioselectivity in radical C−H functionalization of heterocycles. The developed random forest model with physical organic features achieved satisfying performance without any experimental input.