GoFlow: Efficient Transition State Geometry Prediction with Flow Matching and E(3)-Equivariant Neural Networks
Preprint
- 18 June 2025
- preprint
- Published by American Chemical Society (ACS) in ChemRxiv
Abstract
Transition state (TS) geometries of chemical reactions are key to understanding reaction mechanisms and estimating kinetic properties. Inferring these directly from 2D reaction graphs offers chemists a powerful tool for rapid and accessible reaction analysis. Quantum chemical methods for computing TSs are computationally intensive and often infeasible for larger molecular systems. Recently, deep learning–based diffusion models have shown promise in generating TSs from 2D reaction graphs for single-step reactions. However, framing TS generation as a diffusion process, by design, requires a prohibitively large number of sampling steps during inference. Here we show that modeling TS generation as an optimal transport flow problem, solved via E(3)-equivariant flow matching with geometric tensor networks, achieves over a hundredfold speedup in inference while improving geometric accuracy compared to the state-of-the-art. This breakthrough increase in sampling efficiency and predictive accuracy enables the practical use of deep learning-based TS generators in high-throughput settings for larger and more complex chemical systems. Our method, GoFlow, thus represents a significant methodological advancement in machine learning-based TS generation, bringing it closer to widespread use in computational chemistry workflows.Keywords
All Related Versions
Funding Information
- Austrian Science Fund (10.55776/STA192)
- European Research Council (101162908)
This publication has 32 references indexed in Scilit:
- Challenging Reaction Prediction Models to Generalize to Novel ChemistryACS Central Science, 2025
- Accurately Predicting Barrier Heights for Radical Reactions in Solution Using Deep Graph NetworksThe Journal of Physical Chemistry A, 2024
- 3DReact: Geometric Deep Learning for Chemical ReactionsJournal of Chemical Information and Modeling, 2024
- Graph to Activation Energy Models Easily Reach Irreducible Errors but Show Limited TransferabilityThe Journal of Physical Chemistry A, 2024
- Chemprop: A Machine Learning Package for Chemical Property PredictionJournal of Chemical Information and Modeling, 2023
- Best‐Practice DFT Protocols for Basic Molecular Computational Chemistry**Angewandte Chemie International Edition, 2022
- Machine Learning of Reaction Properties via Learned Representations of the Condensed Graph of ReactionJournal of Chemical Information and Modeling, 2021
- Single‐ended transition state finding with the growing string methodJournal of Computational Chemistry, 2015
- Substructural fragments: an universal language to encode reactions, molecular and supramolecular structuresJournal of Computer-Aided Molecular Design, 2005
- The path of chemical reactions - the IRC approachAccounts of Chemical Research, 1981