Molecular field analysis for data-driven molecular design in asymmetric catalysis

Abstract

This review highlights the recent advances in the use of MFA (molecular field analysis) for data-driven catalyst design, enabling to improve selectivity in asymmetric catalysis.

Interplay of Computation and Experiment in Enantioselective Catalysis: Rationalization, Prediction, and─Correction?

“…Their method used Molecular Field Analysis (MFA), which is a regression analysis between the reaction outcomes and molecular fields calculated from 3D structures. Their approach used intermediate structures in the enantiodetermining step to extract and visualize 3D-structural information, similar to other MFA-based methods for predicting stereoselectivity. , The Hong group developed a composite machine learning model that learned from existing stereoselective reactions and was able to accurately and quantitatively predict the activation energies of new reactions . Their composite model outperformed individual models (LASSO, Overall RF, and nucleophile-focused RF) with a mean absolute errors (MAEs) lower across all reaction types.…”

On Accelerating Substrate Optimization Using Computational Gibbs Energy Barriers: A Numerical Consideration Utilizing a Computational Data Set

“…On the other hand, Gibbs energy barriers, calculated using quantum chemical calculations, have been widely used to interpret and explain known reactivities and selectivities of organic reactions. − In addition, their ability to predict results in advance of experiments has also been successful in many examples. ,− In one-step reactions, the Gibbs energy difference between the transition state (TS) and the reactant determines the rate constant and thus represents the reactivity . Even in multistep reactions, the reactivity is represented by the Gibbs energy difference between the TS of the rate-determining step and the lowest intermediate before the rate-determining step, under the generalized pre-equilibrium approximation. , Therefore, this study regards Gibbs energy barriers as experimental and computational reactivities.…”

Using Three-Dimensional Information to Predict and Interpret the Facial Selectivities of Nucleophilic Additions to Cyclic Ketones

“…However, this geometric approach is slightly empirical, and quantitative predictions of facial selectivity when the steric environment of both π-faces is slightly different remain difficult . CoMFA and CoMSIA are based on essential 3D molecular information and are effective for evaluating the steric effects of catalytic ligands . However, these approaches are difficult to apply to substrates with large structural differences, and there are only limited examples of such predictions .…”

“…19 CoMFA 20 and CoMSIA 21 are based on essential 3D molecular information and are effective for evaluating the steric effects of catalytic ligands. 22 However, these approaches are difficult to apply to substrates with large structural differences, and there are only limited examples of such predictions. 23 Although some examples of predictions of the reaction selectivity of B-chlorodiisopinocampheylborane (DIP-Chloride) based on substituent positional information are available, they are limited to the classification of the main products, and no quantitative prediction has yet been reported.…”