ReactionAtlas: Ab origine exploration of chemical reaction networks with machine learning

2026-06-29

AI Summary

Traditional approaches struggle to efficiently construct large-scale chemical reaction networks due to their reliance on prior knowledge and high computational costs. This work proposes a rule-free, automated framework that integrates machine learning force fields (MLFFs) trained on density functional theory (DFT) data with kinetic sampling–based generative models to explore reaction pathways and validate transition states starting from eight prebiotic molecules. For the first time, it enables ab initio construction of a large-scale reaction network incorporating explicit charge and stereochemical information, yielding approximately 47,000 reactions and 12,000 distinct compounds. The MLFF-predicted transition state geometries show excellent agreement with PBE0 reference structures, with 85% exhibiting root-mean-square deviation (RMSD) errors below 0.5 Å, and the approach uncovers novel mechanistic insights into key prebiotic pathways, including the formose cycle.

Abstract

Mapping a chemical reaction network, the graph of minima and transition states (TS) and the elementary reactions connecting them, is the natural language of chemistry, from catalysis to combustion to the origin of life. Constructing such a reaction network for a given chemistry has been impractical: it requires finding and characterizing tens of thousands of TS, a task for which traditional methods such as density functional theory (DFT) are typically prohibitively slow and require reactant and product as input. We introduce ReactionAtlas, which builds a reaction network $\textit{ab origine}$ from a handful of seed molecules and without hand-crafted rules. Specifically, our machine-learned generative model proposes reactions from kinetically sampled candidate compounds and a DFT-trained machine learned force field (MLFF) filters them to valid TS, the resulting products of which enter the search as new seeds. Starting from eight pre-biotic seeds (CH$_2$O, H$_2$O, OH$^-$, H$_3$O$^+$, CO$_2$, H$_2$CO$_3$, HCO$_3^-$, H), ReactionAtlas discovers $\sim$47,000 reactions among $\sim$12,000 compounds. The MLFF TSs match the PBE0 references within 0.5 Å RMSD in 85% of the cases and can be easily brought to the PBE0 level. Thus, ReactionAtlas maps small carbohydrate chemistry up to C$_4$H$_8$O$_4$ at unprecedented scale and accuracy, including charge and stereo information. It enables novel insights into many well-studied reaction paths, including the formose cycle, which we highlight for its centrality to the chemical origins of life. Notably, our framework also allows establishing alternative reaction pathways for formose chemistry.

Problem

• chemical reaction network
• transition states
• ab origine
• prebiotic chemistry
• reaction discovery

Innovation

• Reaction network
• Machine learning force field
• Transition state prediction
• Ab origine exploration
• Prebiotic chemistry