MetaboliteAnnotator: AI-Assisted Name Harmonization and Metadata Enrichment Tool for Metabolomics

Metabolite metadata enrichment remains a significant
challenge
in metabolomics due to the limitations of static databases, incomplete
metabolite coverage, and the labor-intensive nature of manual verification.
Here, we present MetaboliteAnnotator, an R Shiny-based application
for AI-assisted metabolite name harmonization and metadata enrichment.
MetaboliteAnnotator implements a hierarchical procedure, including
preprocessing of input metabolite names, matching against a curated
local resource (covering information on ∼640,000 metabolites
names), PubChem-based real-time retrieval, and AI-assisted matching
for ambiguous compounds, followed by real-time integration of KEGG,
CTD, Reactome, and ChEBI. Compared with MetaboAnalyst 6.0 and MetaboliteIDmapping,
MetaboliteAnnotator achieved significantly higher name hit rates across
all six MetaboLights data sets 93.2% in positive mode (4021/4314 names)
and 93.5% in negative mode (2344/2510 names). MetaboliteAnnotator
outputs standardized identifiers (e.g., InChIKey, PubChem CID), endogenous/exogenous
information, pathway mappings, and metabolite-gene/phenotype associations
for downstream biological interpretation.