Proportion test

Tests whether cell-type proportions differ across groups. CLR-transforms (centered log-ratio) per-sample proportions, then applies a limma-style empirical-Bayes moderated t-test for every pair of groups, with Benjamini–Hochberg FDR applied globally across all (pair × cell-type) hypotheses. Groups can be supplied either via a .obs column (group_col) or via a {sample_id: cluster} mapping (sample_to_clade) from cluster.

Call

from sampledisco.sample_clustering.proportion_test import proportion_test

proportion_test(
    adata=adata_cell,
    sample_col="sample",
    sample_to_clade=expr_clusters,     # from cluster(...)
    celltype_col="cell_type",
    output_dir="sampledisco_demo_output/rna/sample_cluster/expression",
)

Output

Writes → the directory given by output_dir:

• proportion_test_<g1>_vs_<g2>.csv — one file per group pair: per cell-type logFC, p_value, FDR.
• proportion_test_significant_summary.txt — text summary of cell types with FDR < 0.01, per comparison.
• proportion_heatmap_group_by_celltype.png — group-averaged cell-type proportion heatmap (plus CLR-scale and z-score variants).
• proportion_significance_matrix.png — cell-type × comparison significance grid.
• proportion_boxplot_<g1>_vs_<g2>.png — proportion boxplots for the top significant cell types of each comparison.

Result

Per-sample cell-type proportions (left) and which cluster × cell-type combinations are significantly enriched or depleted (right).

See the API page for the full parameter list.