tacco.eval.compute_err

compute_err(adata, result_keys, annotation_keys, err_method='lp', **kwargs)

Computes annotation error of adata’s annotation vs results

Parameters:

• adata – An AnnData including annotations in in .obs or .obsm.

• result_keys – A list of keys in .obsm that contain annotation results

• annotation_keys – The .obs or .obsm keys where the ground truth annotation is stored.

• err_method –

  The method to use to calculate errors. Available are:

  • ”lp”: lp-norm, can use an additional keyword argument ‘p’ with default 2.

  • ”corr”

  • ”max_correct”

  • ”proj”: projection, can use an additional keyword argument ‘metric’ with default “bc”

Returns:

A dict with the errors.