Using Machine Learning to Predict the pKa of C–H Bonds. Relevance to Catalytic Methane Functionalization

Abstract

Six machine learning models (random forest, neural network, support vector machine, k-nearest neighbors, Bayesian ridge regression, least squares linear regression) were trained on a dataset of 3d transition metal-methyl and -methane complexes to predict p<i>K<sub>a</sub></i>(C–H), a property demonstrated to be important in catalytic activity and selectivity. Results illustrate that the machine learning models are quite promising, with RMSE metrics ranging from 4.6 to 8.8 p<i>K<sub>a</sub></i> units, despite the relatively modest amount of data available to train on. Importantly, the machine learning models agreed that (a) conjugate base properties were more impactful than those of the corresponding conjugate acid, and (b) the energy of the highest occupied molecular orbital conjugate base was the most significant input feature in the prediction of p<i>K<sub>a</sub></i>(C–H). Furthermore, results from additional testing conducted using an external dataset of Sc-methyl complexes demonstrated the robustness of all models, with RMSE metrics ranging from 1.5 to 6.6 p<i>K<sub>a</sub></i> units. In all, this research demonstrates the potential of machine learning models in organometallic catalyst development.

Prediction of protein pKawith representation learning

“…Recent studies with machine learning (ML) algorithms for p K a estimations of transition metal complexes have provided new empirical schemes. 89,90 These approaches combine the pattern recognition capabilities of ML algorithms with the atomistic and molecular features that are obtained with a QM tool. However, this scheme can only be practical for proteins if molecular descriptors are obtained with low computational cost, such as neural network potentials (NNPs).…”

Prediction of Protein pKa with Representation Learning

“…Recent studies with machine learning (ML) algorithms for pKa estimations of transition metal complexes provide new empirical schemes 89,90 . These approaches combine the pattern recognition capabilities of ML algorithms with the atomistic and molecular features that are obtained with a QM tool.…”

ChemSpaX: Exploration of Chemical Space by Automated Functionalization of Molecular Scaffold

“…The chemical space is vast and a global exploration is difficult [10,11]. Therefore, machine learning (ML) and other cost-effective computational methods are attractive solutions to navigate the chemical space in the search of novel molecules and materials [12]. A data-driven statistical approach, rooted in quantum and statistical mechanics (QM and SM) is needed to explore and understand the chemical space [13].…”