EnzyKR: A Chirality-Aware Deep Learning Model for Predicting the Outcomes of the Hydrolase-Catalyzed Kinetic Resolution

Hydrolase-catalyzed kinetic resolution is a well-established biocatalytic process. However, the computational tools that predict the favorable enzyme scaffolds for separating racemic mixture are underdeveloped. To address this challenge, we trained a deep learning framework, EnzyKR, to automate the selection of hydrolases for stereoselective biocatalysis. EnzyKR adopts a classifier-regressor architecture that first identifies the reactive binding conformer of an enantiomer-hydrolase complex, and then predicts its activation free energy. A structure-based encoding strategy was used to depict the chiral interactions between hydrolases and enantiomers. EnzyKR was trained using 204 enantiomer-hydrolase complexes curated from IntEnzyDB, and was tested using a pre-split dataset of 20 complexes on the task of active free energy prediction. EnzyKR results in a Pearson R of 0.66, a Spearman R of 0.70, and an MAE of 1.48 kcal/mol. EnzyKR was further tested on the task of predicting enantiomeric excess ratios for 18 hydrolytic reactions catalyzed by fluoroacetate dehalogenase RPA1163 and halohydrin HheC, where the performance of EnzyKR was compared against a recently-developed kinetic predictor, DLKcat. EnzyKR outperformed the DLKcat in 13 out of 18 catalytic reactions. EnzyKR provides a novel computational strategy for an accurate prediction of enantiomeric outcome of hydrolase-catalyzed kinetic resolution reactions.