kaleidocell.get_metaprogram_metrics#

kaleidocell.get_metaprogram_metrics(cluster_dict, mp_dict, similarity_sorted)[source]#

Compute per-meta-program quality metrics.

Parameters:

  • cluster_dict (dict) – {cluster_id: [program_names]}.

  • mp_dict (dict) – {mp_name: pd.Series of gene weights}.

  • similarity_sorted (pd.DataFrame) – Cosine-similarity matrix sorted by cluster assignment (output of derive_nmf_metaprograms()).

Returns:

Rows are meta-program names; columns are: sampleCoverage, silhouette, meanSimilarity, nPrograms, nGenes.

Return type:

pd.DataFrame