scverse / genomic-features

Genomic Features in Python from BioConductor's AnnotationHub
https://genomic-features.readthedocs.io
BSD 3-Clause "New" or "Revised" License
18 stars 5 forks source link

Return better dtypes #46

Open ivirshup opened 1 year ago

ivirshup commented 1 year ago

Description of feature

Right now, we can return some weird dtypes.

Integers with missing values in the ensembl table are cast to float, boolean values are stored as integer, and strings are returned for things better represented as categorical.

We should figure out what dtypes each column should be returned as, and make sure they are returned correctly. It's possible we could pass through pyarrow to do this more efficiently