Public Database Leakage Distorts Model Rankings in Real-Spectra NMR Structure Elucidation
Journal of Chemical Information and Modeling · 12.10.2026 · 66 (19) · 12637-12644
DOI: 10.1021/acs.jcim.6c02552 ↗Kurz gefasst
Fachartikel in der Zeitschrift Journal of Chemical Information and Modeling, online erschienen am 18.09.2026. Verfasst von Zihan Zhang und 4 weiteren Autor:innen. Beteiligte Einrichtung: International College of Pharmaceutical Innovation, Soochow University, Suzhou215123, China. Hauptthemen laut MEDLINE: Databases, Chemical. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 35 Literaturangaben. Gefördert von: Health Research Board, Jiangsu Tepin Foundation.
Abstract (Originalsprache)
Sequence models that translate NMR spectra into molecular structures report up to 96% top-1 accuracy, but they are pretrained on simulated spectra and tested on public databases that can overlap those corpora. We build a leakage-controlled benchmark on nmrshiftdb2 and find that 26.9% of public molecules are exact training-corpus matches under a stated identity rule, with substantial training-set proximity remaining after exact-match removal. On the wider exact-match-removed real-spectrum cohort, released-model top-1 accuracy is 24.4% (95% CI 23.3–25.6%). A training-free rerank using molecular formula and 13C-peak count consistency raises it to 31.1% (29.9–32.4%), supporting leakage-controlled evaluation and decode-time consistency checks before claims of experimental generalization for simulation-pretrained inverse models.
Schlagwörter
MeSH-Schlagwörter (MEDLINE)
Autor:innen und Institutionen
- International College of Pharmaceutical Innovation, Soochow University, Suzhou215123, China.
- School of Pharmacy and Biomolecular Sciences, RCSI, 123 St Stephen's Green, Dublin 2D02 VN51, Ireland.
- Department of Chemistry, RCSI, 123 St Stephen's Green, Dublin 2D02 YN77, Ireland.
ORCID-Profile: Zihan Zhang, Yutao Guo, Dan Wu
Förderung
- Health Research Board
- Jiangsu Tepin Foundation
Zitieren
APA
Zhang, Z., Chen, M., Zhao, X., Guo, Y., & Wu, D. (2026). Public Database Leakage Distorts Model Rankings in Real-Spectra NMR Structure Elucidation. Journal of Chemical Information and Modeling, 66(19), 12637-12644. https://doi.org/10.1021/acs.jcim.6c02552
Vancouver
Zhang Z, Chen M, Zhao X, Guo Y, Wu D. Public Database Leakage Distorts Model Rankings in Real-Spectra NMR Structure Elucidation. Journal of Chemical Information and Modeling. 2026;66(19):12637-12644. doi:10.1021/acs.jcim.6c02552
BibTeX
@article{Zhang2026,
author = {Zhang, Zihan and Chen, Mengxi and Zhao, Xuezhou and Guo, Yutao and Wu, Dan},
title = {Public Database Leakage Distorts Model Rankings in Real-Spectra NMR Structure Elucidation},
journal = {Journal of Chemical Information and Modeling},
year = {2026},
volume = {66},
number = {19},
pages = {12637--12644},
doi = {10.1021/acs.jcim.6c02552}
}
EUVIMED