metachat.pp.generate_adata_met_compass

metachat.pp.generate_adata_met_compass(compass_output, score='Balance', norm='rank', agg='topk_mean', topk=5, **score_kwargs)

Generate processed metabolite matrix for COMPASS analysis using reaction-level penalty scores.

Metabolite-level production / consumption scores are derived from COMPASS reaction penalties via stoichiometric aggregation across all internal reactions that involve each cytoplasmic metabolite. Only metabolites in the [c] compartment are retained and mapped to HMDB IDs.

Reaction metadata (rxn_md.csv) and metabolite annotation (met_md.csv) are loaded from the package’s built-in data.

Parameters

compass_outputstr

Path to the COMPASS output file.

scorestr

Score column to extract for each metabolite. One of "Prod", "Cons", or "Balance" (default). "Balance" is production minus consumption.

normstr

Per-reaction normalisation across cells before aggregation. "rank" (default) uses percentile rank; "minmax" uses min-max scaling.

aggstr

Aggregation method across reactions. "topk_mean" (default) averages the top-k reactions per cell; "max" takes the maximum.

topkint

Number of top reactions used when agg="topk_mean". Default 5.

**score_kwargs

Additional keyword arguments forwarded to _compass_utils.compute_metabolite_scores(), e.g. exclude_transport=False.

Returns

adata_metanndata.AnnData

AnnData with shape (cells × metabolites). obs index = cell IDs; var index = HMDB IDs.