Teaching Language Models Mechanistic Explainability Through Arrow-Pushing

• URL: http://arxiv.org/abs/2512.05722v1
• Date: Fri, 05 Dec 2025 13:57:50 GMT
• Title: Teaching Language Models Mechanistic Explainability Through Arrow-Pushing
• Authors: Théo A. Neukomm, Zlatko Jončev, Philippe Schwaller,
• Abstract summary: Chemical reaction mechanisms provide crucial insight into synthesizability.<n>Current Computer-Assisted Synthesis Planning (CASP) systems lack mechanistic grounding.<n>We introduce a computational framework for teaching language models to predict chemical reaction mechanisms.
• Score: 3.488381738536745
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Chemical reaction mechanisms provide crucial insight into synthesizability, yet current Computer-Assisted Synthesis Planning (CASP) systems lack mechanistic grounding. We introduce a computational framework for teaching language models to predict chemical reaction mechanisms through arrow pushing formalism, a century-old notation that tracks electron flow while respecting conservation laws. We developed MechSMILES, a compact textual format encoding molecular structure and electron flow, and trained language models on four mechanism prediction tasks of increasing complexity using mechanistic reaction datasets, such as mech-USPTO-31k and FlowER. Our models achieve more than 95\% top-3 accuracy on elementary step prediction and scores that surpass 73\% on mech-USPTO-31k, and 93\% on FlowER dataset for the retrieval of complete reaction mechanisms on our hardest task. This mechanistic understanding enables three key applications. First, our models serve as post-hoc validators for CASP systems, filtering chemically implausible transformations. Second, they enable holistic atom-to-atom mapping that tracks all atoms, including hydrogens. Third, they extract catalyst-aware reaction templates that distinguish recycled catalysts from spectator species. By grounding predictions in physically meaningful electron moves that ensure conservation of mass and charge, this work provides a pathway toward more explainable and chemically valid computational synthesis planning, while providing an architecture-agnostic framework for the benchmarking of mechanism prediction.

Related papers

• Agentic reinforcement learning empowers next-generation chemical language models for molecular design and synthesis [51.83339196548892]
  ChemCraft is a novel framework that decouples chemical reasoning from knowledge storage.<n>ChemCraft achieves superior performance with minimal inference costs.<n>This work establishes a cost-effective and privacy-preserving paradigm for AI-aided chemistry.
  arXiv  Detail & Related papers  (2026-01-25T04:23:34Z)
• Foundation Models for Discovery and Exploration in Chemical Space [57.97784111110166]
  MIST is a family of molecular foundation models trained on large unlabeled datasets.<n>We demonstrate the ability of these models to solve real-world problems across chemical space.
  arXiv  Detail & Related papers  (2025-10-20T17:56:01Z)
• oMeBench: Towards Robust Benchmarking of LLMs in Organic Mechanism Elucidation and Reasoning [44.36582860924775]
  We introduce oMeBench, the first large-scale, expert-curated benchmark for organic mechanism reasoning in organic chemistry.<n>We also propose oMeS, a dynamic evaluation framework that combines step-level logic and chemical similarity.
  arXiv  Detail & Related papers  (2025-10-09T03:13:31Z)
• DeepMech: A Machine Learning Framework for Chemical Reaction Mechanism Prediction [2.15242029196761]
  We present DeepMech, an interpretable graph-based deep learning framework to generate chemical reaction mechanisms.<n>DeepMech achieves 98.98 +/-0.12% accuracy in predicting elementary steps and 95.94 +/-0.21% in complete CRM tasks.
  arXiv  Detail & Related papers  (2025-09-19T11:14:46Z)
• Interpretable Deep Learning for Polar Mechanistic Reaction Prediction [43.95903801494905]
  We introduce PMechRP (Polar Mechanistic Reaction Predictor), a system that trains machine learning models on the PMechDB dataset.<n>We train compare a range of machine learning models, including transformer-based, graph-based and two-step siamese architectures.<n>Our best-performing approach was a hybrid model, which combines a 5-ensemble of Chemformer models with a two-step Siamese framework.
  arXiv  Detail & Related papers  (2025-04-22T02:31:23Z)
• Learning Chemical Reaction Representation with Reactant-Product Alignment [50.28123475356234]
  RAlign is a novel chemical reaction representation learning model for various organic reaction-related tasks.<n>By integrating atomic correspondence between reactants and products, our model discerns the molecular transformations that occur during the reaction.<n>We introduce a reaction-center-aware attention mechanism that enables the model to concentrate on key functional groups.
  arXiv  Detail & Related papers  (2024-11-26T17:41:44Z)
• Pre-trained Molecular Language Models with Random Functional Group Masking [54.900360309677794]
  We propose a SMILES-based underlineem Molecular underlineem Language underlineem Model, which randomly masking SMILES subsequences corresponding to specific molecular atoms. This technique aims to compel the model to better infer molecular structures and properties, thus enhancing its predictive capabilities.
  arXiv  Detail & Related papers  (2024-11-03T01:56:15Z)
• A Transformer Based Generative Chemical Language AI Model for Structural Elucidation of Organic Compounds [1.5628118690186594]
  We present a proof-of-concept transformer based generative chemical language artificial intelligence (AI) model. Our model employs an encoder-decoder architecture and self-attention mechanisms to directly generate the most probable chemical structures. It performs structural elucidation of molecules with up to 29 atoms in just a few seconds on a modern CPU, achieving a top-15 accuracy of 83%.
  arXiv  Detail & Related papers  (2024-10-13T15:41:20Z)
• ReactAIvate: A Deep Learning Approach to Predicting Reaction Mechanisms and Unmasking Reactivity Hotspots [4.362338454684645]
  We develop an interpretable attention-based GNN that achieved near-unity and 96% accuracy for reaction step classification. Our model adeptly identifies key atom(s) even from out-of-distribution classes. This generalizabilty allows for the inclusion of new reaction types in a modular fashion, thus will be of value to experts for understanding the reactivity of new molecules.
  arXiv  Detail & Related papers  (2024-07-14T05:53:18Z)
• On the importance of catalyst-adsorbate 3D interactions for relaxed energy predictions [98.70797778496366]
  We investigate whether it is possible to predict a system's relaxed energy in the OC20 dataset while ignoring the relative position of the adsorbate. We find that while removing binding site information impairs accuracy as expected, modified models are able to predict relaxed energies with remarkably decent MAE.
  arXiv  Detail & Related papers  (2023-10-10T14:57:04Z)
• Towards out-of-distribution generalizable predictions of chemical kinetics properties [61.15970601264632]
  Out-Of-Distribution (OOD) kinetic property prediction is required to be generalizable. In this paper, we categorize the OOD kinetic property prediction into three levels (structure, condition, and mechanism) We create comprehensive datasets to benchmark the state-of-the-art ML approaches for reaction prediction in the OOD setting and the state-of-the-art graph OOD methods in kinetics property prediction problems.
  arXiv  Detail & Related papers  (2023-10-04T20:36:41Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.