Towards Generalizable and Evidential Nuclear Magnetic Resonance-Based Molecular Structure Elucidation via Large Language Model Agent
9 affiliations
- 1Information Hub , The Hong Kong University of Science and Technology (Guangzhou) , Guangzhou , China
- 2The University of Hong Kong , Hong Kong , China
- 3State Key Laboratory of Physical Chemistry of Solid Surface , College of Chemistry and Chemical Engineering , Xiamen University , Xiamen , China
- 4Hunan University , Hunan , China
- 5Peking Union Medical College and
- 6The Hong Kong University of Science and Technology , Hong Kong , China
- 7State Key Laboratory of Precision and Intelligent Chemistry , University of Science and Technology of China , Hefei , China
- 8AI for Science Center , Shanghai Artificial Intelligence Laboratory , Shanghai , China
- 9Chinese Academy of Medical Sciences , Beijing , China
Abstract
Nuclear Magnetic Resonance (NMR) spectroscopy is the gold standard for molecular structure elucidation, yet interpreting complex spectra for unknown molecules remains a bottleneck reliant on human expertise. While artificial intelligence has advanced this field, current methods face a critical trade-off: database retrieval cannot identify novel scaffolds, while de novo molecular structure elucidation models operate as black boxes, lacking the atom-level interpretability required for rigorous scientific validation. Here, we present NMRAgent, an evidential reasoning agent powered by large language models (LLMs) that bridges this gap by integrating specialized spectral analysis tools with chemical knowledge graphs. Unlike previous approaches, NMRAgent mimics the deductive reasoning of human experts: it takes experimental NMR spectra and molecular formula as input, plans the elucidation process, proposes candidate structures, verifies peak-atom consistency, and refines misaligned substructure through formula-aware fragment optimization. Enabled by its evidential reasoning, NMRAgent outperforms state-of-the-art methods, improving top-1 accuracy by 46.5% and Tanimoto similarity by 0.502 on a scaffold-split benchmark with novel scaffolds in the test set. Besides, we demonstrate the agent's practical utility by elucidating the structures of two previously unknown natural products isolated from Hydrangea davidii and Vitex trifolia, and by correcting structural misassignments in established literature. By combining high-accuracy prediction with transparent and evidence-based reasoning, NMRAgent establishes a new paradigm for interpretable AI in analytical chemistry.
Related
- MAST: Motif-Augmented Diffusion with Search Tree for Spectroscopic Molecular Structure Elucidation
- Pushing the Limits of One-Dimensional NMR Spectroscopy for Automated Structure Elucidation Using Artificial Intelligence
- Closed-loop Auto Research for Molecular Property Prediction: Discovering and Certifying Generalizable Improvements
- A Modular Rapid-Injection NMR for Structure Elucidation and Kinetic Analysis.
- End-to-end multimodal structure elucidation from raw spectra combining contrastive learning and evolutionary algorithms
BibTeX
@misc{fang2026towards,
title = {{Towards Generalizable and Evidential Nuclear Magnetic Resonance-Based Molecular Structure Elucidation via Large Language Model Agent}},
author = {Zheng Fang and Chen Yang and Yusen Tan and Yunpeng Zhao and Fanjie Xu and Hongxin Xiang and Hanyu Sun and Hanyu Gao and Xiaojian Wang and Wenjie Du and Yuqiang Li and Jun Xia},
howpublished = {arXiv (Cornell University)},
year = {2026},
month = jun,
doi = {10.48550/arxiv.2606.29776},
url = {https://arxiv.org/abs/2606.29776},
note = {CC BY-NC-ND},
}
RIS
TY - UNPB
TI - Towards Generalizable and Evidential Nuclear Magnetic Resonance-Based Molecular Structure Elucidation via Large Language Model Agent
AU - Zheng Fang
AU - Chen Yang
AU - Yusen Tan
AU - Yunpeng Zhao
AU - Fanjie Xu
AU - Hongxin Xiang
AU - Hanyu Sun
AU - Hanyu Gao
AU - Xiaojian Wang
AU - Wenjie Du
AU - Yuqiang Li
AU - Jun Xia
PY - 2026
DA - 2026/06/29
PB - arXiv (Cornell University)
DO - 10.48550/arxiv.2606.29776
UR - https://arxiv.org/abs/2606.29776
AB - Nuclear Magnetic Resonance (NMR) spectroscopy is the gold standard for molecular structure elucidation, yet interpreting complex spectra for unknown molecules remains a bottleneck reliant on human expertise. While artificial intelligence has advanced this field, current methods face a critical trade-off: database retrieval cannot identify novel scaffolds, while de novo molecular structure elucidation models operate as black boxes, lacking the atom-level interpretability required for rigorous scientific validation. Here, we present NMRAgent, an evidential reasoning agent powered by large language models (LLMs) that bridges this gap by integrating specialized spectral analysis tools with chemical knowledge graphs. Unlike previous approaches, NMRAgent mimics the deductive reasoning of human experts: it takes experimental NMR spectra and molecular formula as input, plans the elucidation process, proposes candidate structures, verifies peak-atom consistency, and refines misaligned substructure through formula-aware fragment optimization. Enabled by its evidential reasoning, NMRAgent outperforms state-of-the-art methods, improving top-1 accuracy by 46.5% and Tanimoto similarity by 0.502 on a scaffold-split benchmark with novel scaffolds in the test set. Besides, we demonstrate the agent's practical utility by elucidating the structures of two previously unknown natural products isolated from Hydrangea davidii and Vitex trifolia, and by correcting structural misassignments in established literature. By combining high-accuracy prediction with transparent and evidence-based reasoning, NMRAgent establishes a new paradigm for interpretable AI in analytical chemistry.
SN - 2331-8422
ER -