Enhanced Metabolite Mapping via ChEBI Ontology

Metabolomics datasets often fail to map to Reactome pathways due to systematic differences in chemical specificity between experimental annotations and pathway curation. LC-MS and GC-MS platforms typically report metabolites at a generic level (e.g. undefined stereochemistry or ionization state), whereas Reactome represents pathway participants using highly specific ChEBI identifiers. This mismatch causes many biologically relevant experimental signals to remain unmapped.

Here, we present an ontology-driven workflow that exploits the relational structure of the ChEBI ontology to expand each detected metabolite into a controlled set of chemically equivalent or closely related identifiers. The approach integrates chemical equivalence relationships, limited to a specialization, and ontology-based generalisation to progressively increase pathway coverage while maintaining chemical interpretability.

Applied to multiple public metabolomics datasets, this workflow approximately doubles the number of metabolites mapping to Reactome and yields a two- to three-fold increase in associated pathways compared to RefMet-based mapping alone. The results also highlight chemical classes, particularly lipids, that remain under-represented in pathway annotation, motivating targeted curation and closer integration between metabolomics resources and pathway databases.