Isaac Virshup

Hi @pranzatelli, thanks for reporting this. From the linked issue: > The problem is that the loom file produced by loompy using their reference file for hg38 has a row...

Thanks for the report! As you note, the issue is the index of the dataframe having the same name as one of the columns. I think this is not so...

You should be able to just rename the index. E.g. ```python adata.obs.index.name = "cells" ```

General question about the PR, why buffers specifically and not `h5py.File`/ `h5py.Group` or `zarr` stores directly?

Currently, this works: ```python import anndata as ad from anndata.experimental import read_elem, write_elem import h5py from io import BytesIO adata = ad.read_h5ad("/Users/isaac/data/pbmc3k_raw.h5ad") bio = BytesIO() with h5py.File(bio, "w") as f:...

What is the use case here? I'm not too familiar with Quilt.

For quilt: I saw that, but the specifics were a little unclear. Looks like it's only s3? I think there are some similarities to things I'd like to do with...

One can have backwards compatible pickled objects. I believe both numpy and pandas do this. I don't think it should be too hard for us, as anndata is basically builtin...

~~Could you point me at the deprecated thing in scvi-tools?~~ Found it To me, collapsibility is a must, since the "tons of columns" makes it long quick!

I think this issue will have may subparts, so I think individual issues for that level of specificity I think