Is bioformats2raw .zarr incompatible with ome-zarr-py?

[siemdejong]

Converting an .ome.tiff using bioformats2raw to .zarr and trying to read the .zarr using ome-zarr-py with

from ome_zarr.io import parse_url
from ome_zarr.reader import Reader

url = "bioformats2raw-output.zarr"

reader = Reader(parse_url(url))
# nodes may include images, labels etc
nodes = list(reader())
# first node will be the image pixel data
# IndexError
image_node = nodes[0]

dask_data = image_node.data

errors on image_node = nodes[0]:

---------------------------------------------------------------------------
IndexError                                Traceback (most recent call last)
Cell In[2], [line 10](vscode-notebook-cell:?execution_count=2&line=10)
      [8](vscode-notebook-cell:?execution_count=2&line=8) nodes = list(reader())
      [9](vscode-notebook-cell:?execution_count=2&line=9) # first node will be the image pixel data
---> [10](vscode-notebook-cell:?execution_count=2&line=10) image_node = nodes[0]
     [12](vscode-notebook-cell:?execution_count=2&line=12) # dask_data = image_node.data
     [13](vscode-notebook-cell:?execution_count=2&line=13) # dask_data[0]

IndexError: list index out of range

and indeed, nodes is an empty list.

Are the outputs of bioformats2raw and ome-zarr-py compatible? I tried using an examle OME-TIFF (https://docs.openmicroscopy.org/ome-model/5.6.3/ome-tiff/data.html) which also fails. Currently, bioformats2raw documents it writes OME-NGFF 0.4 by default and ome-zarr-py should currently read 0.4 by default if I'm not mistaken.

Am I missing something?

versions:
bioformats2raw: 0.9.4, git clone https://github.com/glencoesoftware/bioformats2raw && docker build . && docker run ...
ome-zarr-py: 0.10.3

[will-moore]

Not sure what the issue is there. The first code sample is working for me.
What version of zarr do you have there? And fsspec?
I'm using:

ome-zarr==0.10.2
zarr==2.18.2
fsspec==2023.12.1

When you say example OME-TIFF also fails, what did you try there?
Download a sample OME-TIFF and use docker to convert it? What error did you see?

I'm not familiar with using docker for this, so can't help there...

An alternative to setting up the correct environment for bioformats2raw is to use https://www.glencoesoftware.com/products/ngff-converter/ which wraps bioformats2raw in a nice UI.

In either case, the output will be in the bioformats2raw layout, which means that output.zarr contains a series of 1 or more images under output.zarr/0, output.zarr/1 etc. so to read an image with ome_zarr you'll want to do e.g. parse_url("output.zarr/0")

[siemdejong]

Thanks for your quick reply!

I apologize for the confusion. The url in the original comment should have been a local .zarr after OME-TIFF to OME-NGFF conversion with bioformats2raw. Anyway, I wasn't aware that bioformats2raw had a layout that required output.zarr/0 instead of output.zarr. That solved my problem.

[psobolewskiPhD]

I am also puzzled why the bf2raw zarrs don't work quite right, given that there seems to be checking of the metadata:

ome-zarr-py/ome_zarr/utils.py

Line 154 in 5c5b45e

elif zattrs.get("bioformats2raw.layout") == 3:

and then traversal of the xml.

[will-moore]

@psobolewskiPhD Ah - that code is for the $ ome_zarr finder path/to/dir command that walks the file-system, finding ome-zarrs, compiles them into a CSV for BioFile Finder and opens that in the browser.

That code isn't used by parse_url(), although there is a proposal for that at #174