omicverse.metabol.meba

Contents

omicverse.metabol.meba#

omicverse.metabol.meba(adata, *, group_col, time_col, subject_col, groups=None, layer=None)[source]#

Per-feature Hotelling T-squared for two-group time-course comparison.

For each feature, build a (n_subjects, n_timepoints) matrix per group — each row is one subject’s time course. Hotelling’s two-sample T-squared tests whether the two groups have the same mean time-course vector, returning an F-distributed statistic.

The design must be balanced: every subject observed at every time point. Subjects missing any cell are dropped and listed in result.attrs['dropped_subjects'].

Parameters:

  • group_col (str) – Column in adata.obs with the two-class labels.

  • time_col (str) – Column in adata.obs giving the time point (any type castable to str; ordering inferred from pd.unique to preserve appearance order).

  • subject_col (str) – Column in adata.obs identifying each subject. The same subject must appear at every time point in one (and only one) group.

  • groups (Optional[tuple] (default: None)) – (group_a, group_b) pair. None → first two unique values.

  • layer (Optional[str] (default: None)) – AnnData layer (default None → adata.X).

Returns:

Indexed by metabolite, columns T2, F, df1, df2, pvalue, padj, n_a, n_b, k.

Return type:

pd.DataFrame

Contents