LLMs Depend on Specialist Tools for Out-of-Distribution Generalization in MS/MS Structural Elucidation
Abstract
A frontier language model was recently reported to solve 1D NMR structural elucidation without tools. We ask whether this capability transfers to tandem mass spectrometry (MS/MS) in the de novo setting, where the model must propose a structure directly from the spectrum without a candidate list or reference library. To separate reasoning from recall, we build a 100-analyte benchmark from MassSpecGym, split by each analyte's textual footprint into in-distribution (well documented in literature) and out-of-distribution (essentially undocumented) arms. We hold the model, harness, and output format fixed and vary only the LLM's access to domain-specific tools. Without tools, a frontier LLM reaches 52% Top-1 accuracy in-distribution but only 2% out-of-distribution. Only access to de novo structure generators improves out-of-distribution accuracy, raising it to as much as 38% Top-1, and the resulting agent outperforms the standalone de novo models both in and out of distribution. These results suggest that LLMs are not effective at de novo MS/MS structural elucidation on their own, and that further gains depend on the specialist tools available to the agent rather than on the LLM itself.
Related
- In-Context Learning for Multi-Fidelity Multiscale Model Closure and Cross-Operator Out-Of-Distribution Generalization
- Understanding the Out-of-Distribution Generalization of Chain-of-Thought Reasoning in LLMs
- Persistence is not Independence: A structural Credibility Protocol for Distribution based Recovery in VARLiNGAM
- Thick points of log-correlated Gaussian fields do not depend on the mollifier
- Modified treatment policies that depend on the sequence of natural treatment values
BibTeX
@misc{asher2026llms,
title = {{LLMs Depend on Specialist Tools for Out-of-Distribution Generalization in MS/MS Structural Elucidation}},
author = {Gabriel Asher and Cailum M. K. Stienstra and Tomo Oga and Devesh Shah Antonio H. de O. Fonseca and Niall O'Connor and Michael Widrich},
howpublished = {arXiv},
year = {2026},
month = nov,
doi = {10.48550/arxiv.2611.03308},
eprint = {2611.03308},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2611.03308},
}
RIS
TY - UNPB
TI - LLMs Depend on Specialist Tools for Out-of-Distribution Generalization in MS/MS Structural Elucidation
AU - Gabriel Asher
AU - Cailum M. K. Stienstra
AU - Tomo Oga
AU - Devesh Shah Antonio H. de O. Fonseca
AU - Niall O'Connor
AU - Michael Widrich
PY - 2026
DA - 2026/11/02
PB - arXiv
DO - 10.48550/arxiv.2611.03308
UR - https://arxiv.org/abs/2611.03308
AB - A frontier language model was recently reported to solve 1D NMR structural elucidation without tools. We ask whether this capability transfers to tandem mass spectrometry (MS/MS) in the de novo setting, where the model must propose a structure directly from the spectrum without a candidate list or reference library. To separate reasoning from recall, we build a 100-analyte benchmark from MassSpecGym, split by each analyte's textual footprint into in-distribution (well documented in literature) and out-of-distribution (essentially undocumented) arms. We hold the model, harness, and output format fixed and vary only the LLM's access to domain-specific tools. Without tools, a frontier LLM reaches 52% Top-1 accuracy in-distribution but only 2% out-of-distribution. Only access to de novo structure generators improves out-of-distribution accuracy, raising it to as much as 38% Top-1, and the resulting agent outperforms the standalone de novo models both in and out of distribution. These results suggest that LLMs are not effective at de novo MS/MS structural elucidation on their own, and that further gains depend on the specialist tools available to the agent rather than on the LLM itself.
ER -