Count File Format Issue

[npamb]

Hello,

I am having some issues with my count file in the method 1 notebook. I generated my count file by combining the normalized counts from two separate RNAseq datasets and writing that matrix as a h5 file with the rhdf5 package.

When I run

adata = anndata.read_h5ad(counts_file_path)

I get the following error

TypeError                                 Traceback (most recent call last)
File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/utils.py:202, in report_read_key_on_error.<locals>.func_wrapper(*args, **kwargs)
    201 try:
--> 202     return func(*args, **kwargs)
    203 except Exception as e:

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/specs/registry.py:235, in Reader.read_elem(self, elem, modifiers)
    234 if self.callback is not None:
--> 235     return self.callback(read_func, elem.name, elem, iospec=get_spec(elem))
    236 else:

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/h5ad.py:223, in read_h5ad.<locals>.callback(func, elem_name, elem, iospec)
    222 if iospec.encoding_type == "anndata" or elem_name.endswith("/"):
--> 223     return AnnData(
    224         **{
    225             # This is covering up backwards compat in the anndata initializer
    226             # In most cases we should be able to call `func(elen[k])` instead
    227             k: read_dispatched(elem[k], callback)
    228             for k in elem.keys()
    229             if not k.startswith("raw.")
    230         }
    231     )
    232 elif elem_name.startswith("/raw."):

TypeError: __init__() got an unexpected keyword argument 'data'

The above exception was the direct cause of the following exception:

AnnDataReadError                          Traceback (most recent call last)
Cell In[83], line 3
      1 import anndata
----> 3 adata = anndata.read_h5ad(counts_file_path)
      4 adata.shape

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/h5ad.py:243, in read_h5ad(filename, backed, as_sparse, as_sparse_fmt, chunk_size)
    240         return read_dataframe(elem)
    241     return func(elem)
--> 243 adata = read_dispatched(f, callback=callback)
    245 # Backwards compat (should figure out which version)
    246 if "raw.X" in f:

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/experimental/__init__.py:58, in read_dispatched(elem, callback)
     54 from anndata._io.specs import Reader, _REGISTRY
     56 reader = Reader(_REGISTRY, callback=callback)
---> 58 return reader.read_elem(elem)

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/utils.py:204, in report_read_key_on_error.<locals>.func_wrapper(*args, **kwargs)
    202     return func(*args, **kwargs)
    203 except Exception as e:
--> 204     re_raise_error(e, elem)

File ~/mambaforge/envs/cpdb_2/lib/python3.8/site-packages/anndata/_io/utils.py:188, in report_read_key_on_error.<locals>.re_raise_error(e, elem)
    186 else:
    187     parent = _get_parent(elem)
--> 188     raise AnnDataReadError(
    189         f"Above error raised while reading key {elem.name!r} of "
    190         f"type {type(elem)} from {parent}."
    191     ) from e

AnnDataReadError: Above error raised while reading key '/' of type <class 'h5py._hl.files.File'> from /.

I can read my count file with the h5py package. When I check the shape of the data, I get the expected dimensions

f = h5py.File(counts_file_path, 'r')
dset = f['data']
dset.shape

(29966, 14605)

Any suggestions?

Thanks!

[cakirb]

Hello @npamb,

It appears that the rhdf5 package created an .h5 file instead of an .h5ad file, which is likely why the read_h5ad function didn't work as expected. I suggest trying to read the file using the anndata.read_hdf function instead.

Best,
Batu