Beyond RMSD: Deterministic Dlow Matching for Ultrafast and Physically Validated Transition-State Generation
- 1Key Laboratory of Xinjiang Endemic Phytomedicine Resources, Ministry of Education; School of Pharmacy, Shihezi University, Shihezi 832002, China
Abstract
Transition state (TS) generation is typically evaluated by geometric similarity, yet low root-mean-square deviation (RMSD) does not establish first-order saddle-point character or reaction-path connectivity. Here we introduce Kairos-GFM, a deterministic equivariant flow-matching model that combines a two-dimensional reaction graph with the Kabsch-aligned mean of three-dimensional reactant and product geometries. Using 881 Transition1x reactions whose reference TSs passed strict saddle-point and bidirectional intrinsic reaction coordinate validation, Kairos-GFM generates one candidate in 0.0118 s with a median RMSD of 0.14 [A]. A hierarchical physical-validation framework--spanning reaction-center geometry, raw single-point energy, DFT saddle-point optimization, harmonic frequency analysis, and IRC endpoint matching reveals that low RMSD and saddle-point convergence do not guarantee the intended reaction channel. Kairos-GFM recovers the target channel in 75.94% of reactions, outperforming diffusion-based baselines but remaining below transport-based methods. These results position Kairos-GFM as a rapid initial-geometry generator for quantum-chemical refinement and provide a transferable validation standard that distinguishes geometric similarity from physical fidelity.
Related
- Agent-E2MD: Autonomous Translation of Interatomic Potential Equations into Physically Validated Pair Styles for Molecular Dynamics in LAMMPS
- Mid-IR single- and dual-electro-optic comb generation with an ultrafast modulator
- Phonon-Bottleneck-Governed Ultrafast Hot-Carrier Super-Diffusion in Transition Metal Dichalcogenides
- Reinterpreting ultrafast experiments on supercooled water: Glass transition versus liquid-liquid criticality
- Ultrafast electron crystallography reveals the atomic pathway of a light-driven correlated insulator-to-metal transition
BibTeX
@misc{li2026beyond,
title = {{Beyond RMSD: Deterministic Dlow Matching for Ultrafast and Physically Validated Transition-State Generation}},
author = {h. Li and S. Yu and Q. Zhu and Z. Liu and Y. Li and S. Yang},
howpublished = {bioRxiv},
year = {2026},
month = sep,
doi = {10.64898/2026.09.30.755629},
url = {https://www.biorxiv.org/content/10.64898/2026.09.30.755629v1},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Beyond RMSD: Deterministic Dlow Matching for Ultrafast and Physically Validated Transition-State Generation
AU - h. Li
AU - S. Yu
AU - Q. Zhu
AU - Z. Liu
AU - Y. Li
AU - S. Yang
PY - 2026
DA - 2026/09/30
PB - bioRxiv
DO - 10.64898/2026.09.30.755629
UR - https://www.biorxiv.org/content/10.64898/2026.09.30.755629v1
AB - Transition state (TS) generation is typically evaluated by geometric similarity, yet low root-mean-square deviation (RMSD) does not establish first-order saddle-point character or reaction-path connectivity. Here we introduce Kairos-GFM, a deterministic equivariant flow-matching model that combines a two-dimensional reaction graph with the Kabsch-aligned mean of three-dimensional reactant and product geometries. Using 881 Transition1x reactions whose reference TSs passed strict saddle-point and bidirectional intrinsic reaction coordinate validation, Kairos-GFM generates one candidate in 0.0118 s with a median RMSD of 0.14 [A]. A hierarchical physical-validation framework--spanning reaction-center geometry, raw single-point energy, DFT saddle-point optimization, harmonic frequency analysis, and IRC endpoint matching reveals that low RMSD and saddle-point convergence do not guarantee the intended reaction channel. Kairos-GFM recovers the target channel in 75.94% of reactions, outperforming diffusion-based baselines but remaining below transport-based methods. These results position Kairos-GFM as a rapid initial-geometry generator for quantum-chemical refinement and provide a transferable validation standard that distinguishes geometric similarity from physical fidelity.
ER -