Each feature is given a confidence level from its SIRIUS evidence, and its top structure is matched to the exposure compound database (evidence tier, context and human-pathway flag, as for spectra) and to the exposure suspect lists (PubChemLite for Exposomics, NORMAN SusDat, Blood Exposome Database: lists, categories and whether reference MS/MS spectra exist). The Exposure context page gives the source context of the top structures.

Confidence levels (Schymanski et al. 2014) from the SIRIUS evidence alone: 3, tentative structure: the CSI:FingerID top structure with a COSMIC confidence (exact mode) at or above the threshold; 4, molecular formula: the top SIRIUS formula with a ZODIAC score, or else a normalized SIRIUS score, at or above the threshold; 5, exact mass. In silico evidence gives no level 2 or 1: those need a reference spectrum or a reference standard.

Please cite the methods you used: SIRIUS 4 (Dührkop et al., Nat Methods 2019), CSI:FingerID (Dührkop et al., PNAS 2015), COSMIC (Hoffmann et al., Nat Biotechnol 2022), CANOPUS (Dührkop et al., Nat Biotechnol 2021) with ClassyFire (Djoumbou Feunang et al., J Cheminform 2016) and NPClassifier (Kim et al., J Nat Prod 2021), ZODIAC (Ludwig et al., Nat Mach Intell 2020); levels after Schymanski et al., Environ Sci Technol 2014.