recetox-xMSannotator

recetox-xMSannotator annotates LCMS1 features from untargeted high-resolution liquid chromatography–mass spectrometry datasets to assign putative chemical identities and confidence scores using database matching and multistage clustering.

Key Features:

• Database Integration: Matches features against ChemSpider, KEGG, HMDB, T3DB, and LipidMaps to expand candidate chemical annotations.
• Adduct List Usage: Uses a curated adduct list to consider possible ion forms during identification.
• Multistage Clustering Algorithm: Employs multistage clustering incorporating metabolic pathway associations, intensity profiles, retention time characteristics, mass defect, and isotope/adduct patterns.
• Confidence Scoring: Assigns confidence levels to annotations based on integrated evidence from clustering, adduct/isotope patterns, and database matches.
• Performance Metrics: Demonstrates an F1-measure of 0.8 on datasets with known targets and is effective when metabolite counts are small relative to database size.
• MS/MS Validation: MS/MS validation on 210 randomly selected metabolites showed 80% of features annotated with high or medium confidence had ion-dissociation patterns consistent with xMSannotator annotations.

Scientific Applications:

• Untargeted human metabolomics annotation: Annotation and prioritization of features in untargeted human metabolomics datasets using LCMS1 data and database matching.
• Chemical identification in high-resolution MS datasets: Prioritization of putative chemical identities in high-resolution LCMS1 datasets for downstream verification.

Methodology:

Implemented in the R package xMSannotator, the method performs database querying (local and online), applies a curated adduct list, uses multistage clustering based on metabolic pathway associations, intensity profiles, retention time, mass defect, and isotope/adduct patterns, and assigns confidence scores.