Error loading loomxpy formatted files from Fly Cell Atlas

I am trying to load .loom files from https://flycellatlas.org and am getting an odd error. For example:

fly_ann = sc.read_loom('s_fca_biohub_antenna_10x.loom')

gives me

s_fca_biohub_antenna_10x.loom does not appead to be a valid Loom file according to Loom spec version '3.0.0'

error, even though it can be opened in R just fine. Link to the file: https://cloud.flycellatlas.org/index.php/s/PcHBoL23CmxGNsb.

I have a fresh conda environment, scanpy 1.8.1, loompy 3.0.6.

[davidhbrann]

That looks to be an error in loompy from loompy.LoomValidator(). The file has a bunch of row and column attributes that are not part of the loom spec and conventions, which you can find in their documentation here and which you can see in the traceback to ds = loompy.connect('s_fca_biohub_antenna_10x.loom'). You can load the file without validation manually, which worked on my end:

import pandas as pd
import anndata
import loompy

with loompy.connect("s_fca_biohub_antenna_10x.loom", validate=False) as ds:
    print(ds.shape)
    print(ds.ra.keys())
    print(ds.ca.keys())
    genes = ds.ra["Gene"]
    cells = ds.ca["CellID"]
    df = pd.DataFrame(ds[:, :], index=genes, columns=cells).T

adata = anndata.AnnData(df)
print(adata)

Or maybe you can pass the validation=False kwarg to sc.read_loom:

[dkobak]

Thanks a lot!

So setting validation=False kwarg to sc.read_loom does not work: it’s not a supported argument. Setting validate=False resulted in the following error:

ValueError: Cannot construct a Series from an ndarray with compound dtype.  Use DataFrame instead.

But your code snippet worked beautifully!

Do you think it’s something that should be dealt with in scanpy, or should I close this issue and report the problem to the loompy team instead? Or possibly to the Fly Atlas team who created incorrect loom files (I wonder how)?