grassp.pl.annotation_marker_df

annotation_marker_df(data, gt_col, pred_col)

Per-marker probability of that marker’s own compartment.

One row per annotated protein, with the ground-truth label and the probability the annotator gave to that label – the data behind annotation_violin(). Works for any annotator, since the ground-truth one-hot is aligned to the probability matrix by column name.

Return type:

DataFrame

Parameters:

data AnnData

gt_col str

pred_col str