omicverse.metabol.msea_ora

Contents

omicverse.metabol.msea_ora#

omicverse.metabol.msea_ora(hits, background, *, pathways=None, min_size=3, mass_db=None)[source]#

Over-representation analysis via Fisher’s exact test.

Parameters:

  • hits (Iterable[str]) – Metabolite names (e.g. from pyMetabo.significant_metabolites()).

  • background (Iterable[str]) – All tested metabolite names (the universe). Usually adata.var_names after filtering.

  • pathways (Optional[dict[str, list[str]]] (default: None)) – Optional override of {pathway_name: [kegg_id, ...]}. Default is the local KEGG subset shipped with omicverse.

  • min_size (int (default: 3)) – Skip pathways with fewer than this many overlapping background compounds.

  • mass_db (Optional[DataFrame] (default: None)) – Optional pre-fetched ChEBI DataFrame from fetch_chebi_compounds(). When supplied, map_ids uses it as an in-memory lookup for the ~54 k ChEBI names and only falls back to PubChem for names not resolved there. On a cold session this turns the map_ids cost from O(n_features) HTTP round-trips into a single dict probe per feature — often a 30–100x speedup on the first call.

Returns:

Columns: pathway, overlap, set_size, universe_size, odds_ratio, pvalue, padj (BH).

Return type:

pd.DataFrame

Contents