Metabolomics Pathway Enrichment

metabolomics-pathway-enrichment

When to use

The user has a CSV listing metabolites of interest (e.g. significant features from metabolomics-de or metabolomics-statistics, joined with their HMDB / KEGG names) and wants over-representation enrichment via Fisher's exact test, with BH-adjusted FDR.

This is a demo-only enrichment. The pathway database is the hard-coded 9-pathway DEMO_METABOLIC_PATHWAYS dict at met_pathway.py:45-104 (e.g. glycolysis, TCA cycle, amino-acid metabolism). There is NO CLI flag to load real KEGG / Reactome / SMPDB. For production metabolomics enrichment, route to external tools (MetaboAnalystR, mummichog, FELLA) or send the metabolite list through bulkrna-enrichment after gene-mapping.

Inputs & Outputs

Flow

1. Load CSV (--input <metabolites.csv>) or generate a demo at output_dir/<demo>.csv (met_pathway.py:300).
2. Pick the metabolite-list column: metabolite if present, otherwise the first column (met_pathway.py:307).
3. For each pathway in DEMO_METABOLIC_PATHWAYS (met_pathway.py:45), run Fisher's exact test (hypergeometric) (met_pathway.py:132-200); apply BH FDR adjustment (:198).
4. Write tables/pathway_enrichment.csv (met_pathway.py:314) + report.md + result.json.

Gotchas

• Pathway database is HARD-CODED 9 demo pathways. met_pathway.py:45-104 defines DEMO_METABOLIC_PATHWAYS (e.g. glycolysis, TCA cycle, urea cycle). The n_pathways_tested = 9 in result.json (:320) is constant. For real enrichment, use MetaboAnalystR / mummichog / FELLA externally.
• --method mummichog and --method fella are RECORDED-ONLY. met_pathway.py:293 accepts choices=["ora", "mummichog", "fella"] but pathway_enrichment (:132-200) ignores the method parameter — only ORA (Fisher's exact + BH FDR) is implemented. Calling with --method mummichog produces ORA results plus a misleading method=mummichog label in result.json.
• Metabolite-name matching is CASE-INSENSITIVE substring. met_pathway.py:165 lower-cases both query and pathway-member names. glucose, Glucose, D-Glucose all match a pathway entry D-Glucose — but Hexose will NOT.
• Column auto-detection: metabolite first, else first column. met_pathway.py:307 uses met_col = "metabolite" if "metabolite" in df.columns else df.columns[0]. Pre-rename if your CSV has multiple ID columns (name, hmdb_id, kegg).
• --input REQUIRED unless --demo. met_pathway.py:303 raises ValueError("--input required when not using --demo").

Key CLI

# Demo (9-pathway DEMO_METABOLIC_PATHWAYS)
python omicsclaw.py run metabolomics-pathway-enrichment --demo --output /tmp/path_demo

# Real metabolite list (CSV with `metabolite` column)
python omicsclaw.py run metabolomics-pathway-enrichment \
  --input significant_metabolites.csv --output results/

# `--method mummichog` is accepted but produces ORA results regardless
python omicsclaw.py run metabolomics-pathway-enrichment \
  --input significant_metabolites.csv --output results/ \
  --method mummichog

See also

• references/parameters.md — every CLI flag
• references/methodology.md — Fisher's exact ORA, BH FDR, demo-DB caveats
• references/output_contract.md — tables/pathway_enrichment.csv schema
• Adjacent skills: metabolomics-de (upstream — significant feature list), metabolomics-statistics (upstream — multi-test backends), metabolomics-annotation (upstream — m/z → metabolite name mapping), proteomics-enrichment (parallel — same demo-only ORA pattern but for proteins)

Source: TianGzlab/OmicsClaw — distributed by TomeVault.