A large-scale foundation model enables simulation-to-real adaptation for nuclear magnetic resonance-based molecular structure analysis

2026-06-18

AI Summary

This work addresses the scarcity of experimental nuclear magnetic resonance (NMR) data, which has hindered the application of deep learning in molecular structure analysis. To overcome this limitation, the authors propose UltraNMR—the first foundation model that integrates large-scale simulated pretraining with multitask spectral modeling. Leveraging 158 million pairs of simulated ¹H/¹³C NMR spectra, UltraNMR employs contrastive and reconstruction-based pretraining to learn generalizable spectral representations, effectively bridging the gap between simulated and real-world data. The model constructs an NMR vector spectral library encompassing 94 million molecules, enabling structure-aware retrieval. Evaluated on multiple real-world NMR tasks, UltraNMR achieves state-of-the-art performance and successfully elucidates the structures of two unknown natural products listed in the Chinese Pharmacopoeia, significantly outperforming baseline models trained solely on downstream experimental data.

Abstract

Nuclear Magnetic Resonance (NMR) spectroscopy is a powerful tool for molecular structure analysis, and spectral artificial intelligence offers great potential for its rapid and automated interpretation. However, the scarcity of experimental NMR datasets has constrained deep learning in this domain to narrow, task-specific applications that lack broad generalization. Here, we introduce UltraNMR, a large-scale foundation model for NMR that leverages the intrinsic properties of NMR spectra to learn generalizable spectral representations. We collected 158 million paired simulated $^{1}$H and $^{13}$C NMR spectra to train UltraNMR, employing multiple domain-specific pre-training objectives. UltraNMR captures both intra-spectral and inter-spectral dependencies, enabling seamless simulation-to-real adaptation. We demonstrate that adapting UltraNMR to a range of molecular structure analysis tasks on experimental NMR spectra consistently yields state-of-the-art performance and clearly outperforms UltraNMR variants trained directly on downstream data without simulation pre-training. We also construct a large-scale NMR spectral vector library by encoding simulated NMR spectra using UltraNMR, covering 94 million unique molecules and enabling effective structure-aware retrieval. In real-world applications, UltraNMR facilitates the structural elucidation of two previously unknown natural products from Chinese herbal medicines recorded in the Chinese Pharmacopoeia. These results suggest that large-scale simulation pre-training can effectively bridge the simulation-to-real gap, enabling robust and generalizable molecular structure analysis of real-world NMR spectra.

Problem

• NMR spectroscopy
• data scarcity
• simulation-to-real gap
• molecular structure analysis
• generalization

Innovation

• foundation model
• simulation-to-real adaptation
• NMR spectroscopy
• large-scale pre-training
• molecular structure analysis