Prediction of transition state structures of gas-phase chemical reactions via machine learning

Abstract

The elucidation of transition state (TS) structures is essential for understanding the mechanisms of chemical reactions and exploring reaction networks. Despite significant advances in computational approaches, TS searching remains a challenging problem owing to the difficulty of constructing an initial structure and heavy computational costs. In this paper, a machine learning (ML) model for predicting the TS structures of general organic reactions is proposed. The proposed model derives the interatomic distances of a TS structure from atomic pair features reflecting reactant, product, and linearly interpolated structures. The model exhibits excellent accuracy, particularly for atomic pairs in which bond formation or breakage occurs. The predicted TS structures yield a high success ratio (93.8%) for quantum chemical saddle point optimizations, and 88.8% of the optimization results have energy errors of less than 0.1 kcal mol−1. Additionally, as a proof of concept, the exploration of multiple reaction paths of an organic reaction is demonstrated based on ML inferences. I envision that the proposed approach will aid in the construction of initial geometries for TS optimization and reaction path exploration.

Atomic fragment approximation from a tensor network

“…The intermediate prediction accuracy of AFA is 94.34%, while a recent model has an accuracy of 93.8%. 34 Here the accuracy is measured as top-2 accuracy, where AFA provides the top two most likely intermediate candidates; if the correct intermediate has the same structure as one of them, we consider the prediction to be accurate. Generally speaking, small effects may affect the possible reaction pathways, so further validation and refinement are required to clearly identify the reaction pathway.…”

Diffusion-based Generative AI for Exploring Transition States from 2D Molecular Graphs

“…The COV score measures the percentage of the reference TS geometries covered by the predicted ones by TSDiff, where a reference is considered to be covered if there exists any predicted one having a D-MAE of 0.1 Å or less with the reference. This criterion of 0.1 Å was determined based on the accuracy of a state-of-the-art model [52] which has demonstrated reliability with a high success rate in quantum chemical validations. The MAT score measures the similarity between generated and reference samples by calculating the minimum D-MAE between the generated geometries and the reference geometry.…”

CoeffNet: predicting activation barriers through a chemically-interpretable, equivariant and physically constrained graph neural network

“…More complex models consider other potential energy surface (PES) complexities, such as solvent reorganization in Marcus theory. 17–21 While these models work well for a narrow chemical space and are chemically interpretable, they need to be re-parameterized for even the slightest change to the chemistry and conditions, such as for example, different ligands around a fixed reaction centre.…”