Towards Generalizable and Evidential Nuclear Magnetic Resonance-Based Molecular Structure Elucidation via Large Language Model Agent

2026-06-29

AI Summary

This work addresses the challenges in nuclear magnetic resonance (NMR) spectral interpretation—namely, the heavy reliance on expert knowledge and the inability of existing AI methods to simultaneously handle novel molecular scaffolds and provide atom-level interpretability. To this end, we propose NMRAgent, a large language model–based agent that integrates a chemical knowledge graph, specialized spectral analysis tools, and a formula-aware fragment optimization algorithm. NMRAgent introduces, for the first time, an evidence-based reasoning mechanism that emulates human experts’ deductive processes to derive, validate, and refine molecular structures from NMR spectra and molecular formulas, while delivering transparent and verifiable atom-level reasoning traces. On a test set containing novel scaffolds, our method achieves a 46.5% improvement in Top-1 accuracy and a 0.502 increase in Tanimoto similarity, successfully elucidating the structures of two new natural products and correcting misassignments reported in the literature.

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.

Problem

• NMR spectroscopy
• molecular structure elucidation
• novel scaffolds
• interpretability
• evidential reasoning

Innovation

• evidential reasoning
• large language model agent
• NMR-based structure elucidation
• interpretable AI
• de novo molecular design