核磁共振(NMR)光谱学是分子结构解析的基础,但大规模解读光谱仍耗时且高度依赖专业知识。尽管最近的光谱作为语言建模和基于检索的方法显示出前景,但它们严重依赖于大规模计算光谱的语料库,并在应用于实验测量时表现出显著的性能下降。为了解决这些问题,我们构建了NMRSpec,这是一个从化学文献中挖掘出的大量实验¹H和¹³C光谱的语料库,并提出了NMRTrans,它将光谱建模为无序的峰集,并将模型的归纳偏置与NMR的物理性质对齐。据我们所知,NMRTrans是首个仅基于大规模实验光谱训练的NMR Transformer,并在实验基准上实现了最先进的性能,其Top-10准确率比最强基线提高了+17.82点(61.15% vs. 43.33%),强调了实验数据和结构感知架构对可靠NMR结构解析的重要性。
英文摘要:
Nuclear Magnetic Resonance (NMR) spectroscopy is fundamental for molecular structure elucidation, yet interpreting spectra at scale remains time-consuming and highly expertise-dependent. While recent spectrum-as-language modeling and retrieval-based methods have shown promise, they rely heavily on large corpora of computed spectra and exhibit notable performance drops when applied to experimental measurements. To address these issues, we build NMRSpec, a large-scale corpus of experimental $^1$H and $^{13}$C spectra mined from chemical literature, and propose NMRTrans, which models spectra as unordered peak sets and aligns the model's inductive bias with the physical nature of NMR. To our best knowledge, NMRTrans is the first NMR Transformer trained solely on large-scale experimental spectra and achieves state-of-the-art performance on experimental benchmarks, improving Top-10 Accuracy over the strongest baseline by +17.82 points (61.15% vs. 43.33%), and underscoring the importance of experimental data and structure-aware architectures for reliable NMR structure elucidation.
NMR-Solver: Automated Structure Elucidation via Large-Scale Spectral Matching and Physics-Guided Fragment Optimization
School of Mathematical Sciences(数学科学学院);Peking University(北京大学);DP Technology(DP科技);College of Chemistry and Molecular Engineering(化学与分子工程学院);Xiamen University(厦门大学);AI for Science Institute(人工智能科学研究院);Center for Machine Learning Research, Peking University(北京大学机器学习研究中心)
Multi-Agent Closed-Loop Reasoning for Organic Structure Elucidation from Multimodal Spectra
Shanghai Jiao Tong University(上海交通大学);Institute of Process Engineering, Chinese Academy of Sciences(中国科学院过程工程研究所);Huashan Hospital, Fudan University(复旦大学附属华山医院);Minhang District Center for Disease Control(闵行区疾病预防控制中心);Johns Hopkins University(约翰斯·霍普金斯大学);National Clinical Research Center for Kidney Diseases(国家肾脏疾病临床医学研究中心)
Direct deduction of chemical class from NMR spectra
University of Tartu(塔尔图大学);De Montfort University(德蒙福特大学);Mestrelab Research, S.L.(Mestrelab研究有限公司);University of Limerick(利默里克大学);Swansea University(斯旺西大学);Universidade Federal do Rio de Janeiro(里约热内卢联邦大学)
TransPeakNet: Solvent-Aware 2D NMR Prediction via Multi-Task Pre-Training and Unsupervised Learning
Brandeis University(布兰迪斯大学);Harvard Medical School(哈佛医学院);Complex Carbohydrate Research Center, University of Georgia(佐治亚大学复合碳水化合物研究中心);Boston College(波士顿学院)
A large-scale foundation model enables simulation-to-real adaptation for nuclear magnetic resonance-based molecular structure analysis
The Hong Kong University of Science and Technology(香港科学与技术大学);Peking Union Medical College(北京友谊医院);Chinese Academy of Medical Sciences(中国医学科学院);Xiamen University(厦门大学);Hunan University(湖南大学);Shanghai Artificial Intelligence Laboratory(上海人工智能实验室)
MolP2SBench: Do LLMs Really Reason from Molecular Properties to Structure?
University of Tennessee, Knoxville(田纳西大学诺克斯维尔分校);University of Illinois at Urbana-Champaign(伊利诺伊大学厄巴纳-香槟分校);University of California, Berkeley(加州大学伯克利分校);University of Georgia(佐治亚大学);Florida State University(佛罗里达州立大学)
Hypothesis-and-Refinement Learning of Organic Structures from Multimodal Spectroscopic Data
State Key Laboratory of Advanced Waterproof Materials, School of Materials Science and Engineering, Peking University(先进防水材料国家重点实验室,材料科学与工程学院,北京大学);School of Electronic and Computer Engineering, Peking University Shenzhen Graduate School(电子与计算机工程学院,北京大学深圳研究生院);Peng Cheng Laboratory(鹏城实验室);IBS Center for Algorithmic and Robotized Synthesis (CARS), UNIST(算法与机器人化合成中心(CARS),UNIST);Department of Chemistry, UNIST(化学系,UNIST);School of Advanced Materials, Peking University Shenzhen Graduate School(先进材料学院,北京大学深圳研究生院)