A unified machine-learning protocol for asymmetric catalysis as a proof of concept demonstration using asymmetric hydrogenation

Abstract

Design of asymmetric catalysts generally involves time- and resource-intensive heuristic endeavors. In view of the steady increase in interest toward efficient catalytic asymmetric reactions and the rapid growth in the field of machine learning (ML) in recent years, we envisaged dovetailing these two important domains. We selected a set of quantum chemically derived molecular descriptors from five different asymmetric binaphthyl-derived catalyst families with the propensity to impact the enantioselectivity of asymmetric hydrogenation of alkenes and imines. The predictive power of the random forest (RF) built using the molecular parameters of a set of 368 substrate–catalyst combinations is found to be impressive, with a root-mean-square error (rmse) in the predicted enantiomeric excess (%ee) of about 8.4 ± 1.8 compared to the experimentally known values. The accuracy of RF is found to be superior to other ML methods such as convolutional neural network, decision tree, and eXtreme gradient boosting as well as stepwise linear regression. The proposed method is expected to provide a leap forward in the design of catalysts for asymmetric transformations.

Regio-selectivity prediction with a machine-learned reaction representation and on-the-fly quantum mechanical descriptors

“…The recent advance in high-throughput experimentation and data-mining techniques and thus the presence of high-quality data, have signicantly populated ML methods in chemical reactivity predictions. [20][21][22][23][24][25][26][27][28] Recently, Sigman and co-workers 21,22 advanced multivariate linear regression to predict the selectivity of a reaction (formally, the difference of free energy barriers), by relying on sophisticated electronic and steric descriptors of substrates and catalysts. An alternate statistical approach built on support vector machines (SVM) and feed-forward neural networks (FFNN) was demonstrated by Denmark and coworkers, 23 in which the authors proposed a new 3D shape descriptor for catalysts, the average steric occupancy (ASO).…”

A review on extreme learning machine

“…Ouyang and Wang [129] applied ELM into measuring NOx in vehicle exhaust and achieved qualified experiment results. And it could be observed that more scientists were still putting efforts into applying ELM based methods to chemical analysis [162,180].…”

Highly accurate prediction of reaction performance and enantioselectivity with a uniform machine learning protocol

“…Historically, the problem was driven by chemical intuition and empiricism with enormous experimental effort. Recently, emergence of machine learning approaches to facilitate synthetic chemistry and asymmetric catalysis has provided an attractive alternative to this long-standing challenge [3][4][5][6] . First, without the need to consider reaction mechanism, quantifying molecular properties derived from various reaction substrates and catalysts can be characterized by designed multidimensional descriptors in a certain type reaction.…”