You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
johnkerl
changed the title
[python] to_anndata inconsistency in handling of obsp/varp
[python] Inconsistency in to_anndata handling of obsp/varpNov 21, 2024
johnkerl
changed the title
[python] Inconsistency in to_anndata handling of obsp/varp
[python] Inconsistency in to_anndata density/sparsity of obsp/varpNov 22, 2024
Oh let me clarify, it's not that it should be sparse in all cases. It's just that in the majority of cases it should be sparse. If the user really needs a dense array (maybe for a correlation matrix), they can do that.
Ideally you support both. I think the user would decide which they want by what format they stored it in.
While
tiledbsoma.io.to_anndata()
exportsobsp
/varp
as ascipy.sparse
matrix,ExperimentAxisQuery.to_anndata()
exports it as a dense ndarray.The AnnData API suggests the dense ndarray.
Update on #3362: @ivirshup points out that the AnnData docs are wrong and these should be sparse in all cases.
[sc-59632]
The text was updated successfully, but these errors were encountered: