Natural-Product Spectra Need Models That Know When They Are Outside Domain: An Uncertainty-Calibrated Architecture for MS/MS- and NMR-Assisted Structure Prioritization
Year: 2026
Type: article
Abstract: Natural-product structure prioritization increasingly uses machine-learning models that map tandem mass spectra, nuclear magnetic resonance data, or combinations of spectral modalities to candidate structures. Yet a high model score does not establish that a query lies within the chemistry, acquisition conditions, spectral information regime, or candidate space represented during model development. This distinction matters especially for natural products, where scaffold novelty, sparse reference
DOI: 10.51847/tn3lzglxg6
Language: English
Subfield: Drug Discovery
Domain: Life Sciences
Open Access status: bronze
OpenAlex ID: W7204993905
