Metabolite Annotation Using MS2 Fragmentation

Peaks were classified and tentatively annotated with SIRIUS (Dührkop et al., 2019 (link)), ZODIAC (Ludwig et al., 2020 ), CSI:FingerID (Dührkop et al., 2015 ; Hoffmann et al., 2021 ), and CANOPUS (Djoumbou Feunang et al., 2016 (link); Dührkop et al., 2021 (link)) using ClassyFire and NPClassifier (Djoumbou Feunang et al., 2016 (link); Kim et al., 2021 (link)) based on MS2 fragmentation patterns using SIRIUS software version 5.6.2. Default settings were used for SIRIUS, ZODIAC, and CSI:FingerID, however only formulas from natural-product based databases were considered (Bio Datadase, Biocyc, CHEBI, COCONUT, EcoCyc Mine, CNPS, HMDB, KEGG, KEGG Mine, KNApSAcK, Natural Products and Plantcyc) for CANOPUS in the SIRIUS software suite. To tentatively identify a compound, a combination of COSMIC and ZODIAC scores were considered. Manual analysis of the matching substructures was conducted before structural assignment was made. If the fragmentation pattern did not match the structures proposed by CSI:FingerID, matching fragments from the proposed structures were considered when the class was assigned using CANOPUS. If the ZODIAC score was < 50%, no tentative identification was made. If the SIRIUS score was < 50% with no accompanying ZODIAC score, no identification was made. For an overview of the detected compound classes, a sunburst plot of all classified features was constructed.