Multi-Instance Learning Approach to Predictive Modeling of Catalysts Enantioselectivity

Here, we report an application of the multi-instance learning approach to predictive modeling of enantioselectivity of chiral catalysts. Catalysts were represented by ensembles of conformations encoded by the pmapper physicochemical descriptors capturing stereo-configuration of the molecule. Each catalyzed chemical reaction was transformed to a condensed graph of reaction for which ISIDA fragment descriptors were generated. This approach does not require any conformations' alignment and can potentially be used for a diverse set of catalysts bearing different scaffolds. Its efficiency has been demonstrated in predicting the selectivity of BINOL-derived phosphoric acid catalysts in asymmetric thiol addition to N-acylimines and benchmarked with previously reported models.

Automated summary

This study applies a multi-instance learning approach to predictive modeling of enantioselectivity of chiral catalysts, representing catalysts with pmapper physicochemical descriptors and generating ISIDA fragment descriptors for catalyzed chemical reactions, without the need for conformation alignment. The method efficiently predicts the selectivity of BINOL-derived phosphoric acid catalysts in asymmetric reactions and is benchmarked against previously reported models.