Abstract:Activation barriers of elementary reactions are essential to predict molecular reaction mechanisms and kinetics. However, computing these energy barriers by identifying transition states with electronic structure methods ( e.g. , density functional theory) can be time-consuming and computationally expensive. In this work, we introduce CoeffNet , an equivariant graph neural network that predicts activation barriers using coefficients of any frontier molecular orbital (such as the highest occupied molecular orbital) of reactant and product complexes as graph node features. We show that using coefficients as features offer several advantages, such as chemical interpretability and physical constraints on the network's behaviour and numerical range. Model outputs are either activation barriers or coefficients of the chosen molecular orbital of the transition state; the latter quantity allows us to interpret the results of the neural network through chemical intuition. We test CoeffNet on a dataset of S N 2 reactions as a proof-of-concept and show that the activation barriers are predicted with a mean absolute error of less than 0.025 eV. The highest occupied molecular orbital of the transition state is visualized and the distribution of the orbital densities of the transition states is described for a few prototype S N 2 reactions.

What problem does this paper attempt to address?

### What problems does this paper attempt to solve? This paper aims to solve the challenges faced when calculating the activation barriers of reactions. Specifically, it is crucial to predict molecular reaction mechanisms and kinetics by identifying transition states, but calculating these energy barriers using electronic structure methods (such as density functional theory, DFT) is usually time - consuming and computationally expensive. To solve this problem, the author introduced a new model - **CoeﬀNet**, which is an equivariant graph neural network. It uses the coefficients of any frontier molecular orbitals (such as the highest occupied molecular orbital, HOMO) of reactant and product complexes as graph node features to predict the activation barriers. This method not only improves the prediction accuracy but also provides chemical interpretability and imposes physical constraints to ensure the rationality of the model behavior and numerical range. #### Main objectives: 1. **Improve computational efficiency**: Compared with traditional DFT calculations, CoeﬀNet can predict activation barriers more quickly. 2. **Provide chemical interpretability**: By using molecular orbital coefficients as features, the model's output can be interpreted based on chemical intuition. 3. **Impose physical constraints**: Ensure that the model learns within a fixed numerical range, thereby improving the reliability of prediction. #### Specific applications: - **SN2 reaction**: As a proof of concept, the author tested the performance of CoeﬀNet on a set of SN2 reactions, and the results showed that the mean absolute error was less than 0.025 eV. - **Visualization and analysis**: The model can also predict and visualize the highest occupied molecular orbital (HOMO) of the transition state and describe its orbital density distribution. Through these improvements, CoeﬀNet can not only accurately predict activation barriers but also provide valuable chemical explanations for chemists, thereby guiding further research and experimental design.

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

Accurately estimating activation energies by leveraging neural network methods and a large dataset

Enhancing Activation Energy Predictions under Data Constraints Using Graph Neural Networks

Precise estimation of activation energies in gas-phase chemical reactions via artificial neural network

Towards the design of chemical reactions: Machine learning barriers of competing mechanisms in reactant space

CatTSunami: Accelerating Transition State Energy Calculations with Pre-trained Graph Neural Networks

A Neural Network Protocol for Predicting Molecular Bond Energy

Graph to Activation Energy Models Easily Reach Irreducible Errors but Show Limited Transferability

The importance of reaction energy in predicting chemical reaction barriers with machine learning models

InteractionNet: Modeling and Explaining of Noncovalent Protein-Ligand Interactions with Noncovalent Graph Neural Network and Layer-Wise Relevance Propagation

Accurately Predicting Barrier Heights for Radical Reactions in Solution Using Deep Graph Networks

Mechanism Deduction from Noisy Chemical Reaction Networks

Modeling protein-small molecule conformational ensembles with ChemNet

Multi-Task Mixture Density Graph Neural Networks for Predicting Cu-based Single-Atom Alloy Catalysts for CO2 Reduction Reaction

Nucleophilicity Prediction Using Graph Neural Networks

Graph Nets for Partial Charge Prediction

Multi‐Task Mixture Density Graph Neural Networks for Predicting Catalyst Performance

KineticNet: Deep learning a transferable kinetic energy functional for orbital-free density functional theory

Predicting Organic Reaction Outcomes with Weisfeiler-Lehman Network

Neural Network Potential with Multi-Resolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution