tacco.preprocessing.subsample_annotation

subsample_annotation(adata, annotation_key=None, modification=None, range_factor=1, seed=42)[source]

Subsamples the observations to change the annotation fractions.

Parameters:

  • adata – An AnnData containing categorical annotation to subsample.

  • annotation_key – The .obs key with the categorical annotation. Is determined automatically if possible.

  • modification – A dict-like mapping the annotation categories to a float giving the fraction of this category to be kept. Unlisted categories are unchanged. Instead of the full name of a category, also unambiguous shortcuts are possible, i.e. the first few letters.

  • range_factor – A float giving a factor to scale down all categories a priori. A factor larger than 1 (e.g. 2) reduces all category observation and makes values larger than 1 in modification possible without duplicated observations.

  • seed – The random seed to use

Returns:

Returns a Series with the random recaling factors.