Remove samples/rows from anndata

Help anndata
1

I can see how to remove columns from anndata ie

keep = ['a','b','c']

adata = adata [:, keep]

How does one remove rows/sets of samples from anndata.obs and anndata.x?

for example remove adata.obs[Region='reg012']

adata.obs

enter image description here

2

Everything is connected, so you can’t remove rows from .obs without removing rows from the whole object. In your case you want to do one of the following:

adata[adata.obs["Region"] != "reg012"]
adata[adata.obs.query("Region != 'reg012'").index]
adata[[r != "reg012" for r in adata.obs.Region]]
3 Likes
3

@adamgayoso thanks.
How does one do the inverse of keep for adata.var_names?
ie to exclude based on input list rather than include.
exclude = ['A','B','C']
àdata = adata[:, exclude]

4 
adata = adata[:, ~adata.var_names.isin(exclude)]

Tilde is “not”, so this reads like “var names that are not in exclude”

1 Like
5

Hi there. If I have a list of cluster labels. And I want to remove all the cells in my anndata that have a label that belongs to said list called clusters_to_remove. How can I go about doing that? @adamgayoso

6

clusterstoremove= [‘1’,‘2’,‘3’]
adata = adata[adata.obs.clustercolumname.isin(clusterstoremove)==FALSE].copy()

1 Like