Unable to edit backed AnnData

[jpmcga]

Hi,

Thanks again for writing a wonderful package. I'm currently trying to implement the backed AnnData objects into a pipeline I'm building. I have a couple items of feedback to share, but I'll start with this one.

After I create backed AnnData objects via pp.import_data I'm unable to add new columns to .obs (i.e., adata.obs["barcode"] = adata.obs_names). Attempting to do so, I get the error

called `Result::unwrap()` on an `Err` value: H5Ldelete(): unable to delete link: no write intent on file note: run with `RUST_BACKTRACE=1` environment variable to display a backtrace

I'm able to resolve this error by closing the AnnData object (adata.close()), and then rereading with write permissions (adata = snap.read(f"{file_name}.h5ad", "r+")). While this works, it's awkward and I have to do it again after running snap.metrics.tsse. I wonder if there's a better way to specify permissions when create backed AnnData objects. If not, perhaps an option to specify in import_data would be helpful.

Thanks!

[kaizhang]

This is wired. The following code works for me:

import snapatac2 as snap

fragment_file = snap.datasets.pbmc500(downsample=True)
data = snap.pp.import_data(
    fragment_file,
    chrom_sizes=snap.genome.hg38,
    file='test.h5ad',
    sorted_by_barcode=False,
)
data.obs['barcode'] = data.obs_names

print(data)

[jpmcga]

Thanks for the quick response! The code you shared works for me as well. However, I'm able to trigger the error by putting fragment_file and the file name in lists:

import snapatac2 as snap

if __name__ == '__main__':

    fragment_file = snap.datasets.pbmc500(downsample=True)
    data = snap.pp.import_data(
        [fragment_file],
        chrom_sizes=snap.genome.hg38,
        file=['test.h5ad'],
        sorted_by_barcode=False,
    )
    data[0].obs['barcode'] = data[0].obs_names

    print(data)

Any ideas? Thanks.