A Priori Sampling of Transition States with Guided Diffusion

2026-03-26

AI Summary

This work proposes a novel paradigm for transition state search that eliminates the need for heuristic initial guesses or prior knowledge of reaction coordinates. By training a score-based diffusion model on metastable configurations and integrating conditional score guidance with a Score-Aligned Ascent (SAA) algorithm, the method steers the inference trajectory toward the equidensity manifold separating metastable basins, followed by physics-driven refinement to precisely converge to first-order saddle points. This approach enables, for the first time, generative-model-based sampling of multiple reaction pathways. It achieves high-accuracy transition state localization and uncovers diverse reaction channels across benchmark systems, including two-dimensional potential energy surfaces, biomolecular conformational transitions, and chemical reactions.

Abstract

Transition states, the first-order saddle points on the potential energy surfaces, govern the kinetics and mechanisms of chemical reactions and conformational changes. Locating them is challenging because transition pathways are topologically complex and can proceed via an ensemble of diverse routes. Existing methods address these challenges by introducing heuristic assumptions about the pathway or reaction coordinates, which limits their applicability when a good initial guess is unavailable or when the guess precludes alternative, potentially relevant pathways. We propose to bypass such heuristic limitations by introducing ASTRA, A Priori Sampling of TRAnsition States with Guided Diffusion, which reframes the transition state search as an inference-time scaling problem for generative models. ASTRA trains a score-based diffusion model on configurations from known metastable states. Then, ASTRA guides inference toward the isodensity surface separating the basins of metastable states via a principled composition of conditional scores. A Score-Aligned Ascent (SAA) process then approximates a reaction coordinate from the difference between conditioned scores and combines it with physical forces to drive convergence onto first-order transition states. Validated on benchmarks from 2D potentials to biomolecular conformational changes and chemical reaction, ASTRA locates transition states with high precision and discovers multiple reaction pathways, enabling mechanistic studies of complex molecular systems.

Problem

• transition states
• reaction pathways
• potential energy surfaces
• conformational changes
• chemical reactions

Innovation

• guided diffusion
• transition state search
• score-based generative model
• reaction pathway discovery
• Score-Aligned Ascent