Support for cells.zarr.zip for Xenium

[LucaMarconato]

Depends on #111 (should be merged first).

Closes #110

• reading the labels for nuclei and cells
• reading the columns of cells_summary and adding it to the table

[codecov-commenter]

Codecov Report

Attention: Patch coverage is 25.53191% with 70 lines in your changes are missing coverage. Please review.

Project coverage is 35.94%. Comparing base (f0d1f45) to head (a2e49fc).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #112      +/-   ##
==========================================
- Coverage   36.33%   35.94%   -0.39%     
==========================================
  Files          18       18              
  Lines        1434     1516      +82     
==========================================
+ Hits          521      545      +24     
- Misses        913      971      +58     

Files	Coverage Δ
src/spatialdata_io/_constants/_constants.py	100.00% <100.00%> (ø)
src/spatialdata_io/readers/xenium.py	22.68% <23.07%> (+0.98%)	⬆️

... and 1 file with indirect coverage changes

[LucaMarconato]

Tested on all the public Xenium versions using this https://github.com/giovp/spatialdata-sandbox/blob/main/tests/test_to_zarr_xenium.py

[mattcai10x]

Not all v2.0.0 data has 4 channels. Xenium customers can run samples without the cell segmentation kit, in which case they will only have 1 ome.tif file in the morphology_focus_dir.

[LucaMarconato]

Thanks for the clarification, I will change the code to expect either the single DAPI channel either all the 4 channels.

If in the future also other cases are supported, such as an intermediate number of channels, I will adapt the code and make it general.

[LucaMarconato]

I applied the code suggestions and performed some additional cleanup. Also, I realized that the link between the labels indices for the cell segmentation and the instances of the rows required to parse additional data from cells.zarr.zip. Now this is included.

I tested the code on the Xenium Analyzer versions 1.0.1, 1.0.2, 1.3.0, 1.4.0, 1.5.0, 1.6.0, 1.7.0, 2.0.0.

One note on the new labels: the table is still mapped to the circles/shapes but the user can change the annotation target of the table to the labels using the new set_table_annotates_spatialelement() method. I plan to show this in an example notebook.

Using the multiscale labels instead of the shapes for visualizing the data not only improves the quality of the visualization over circles and polygons, but also doesn't compromise performance. Here is an example of the new (labels) vs old (circles) napari view.

Full code to load the data, change the table and view the data.

##
import spatialdata as sd
from napari_spatialdata import Interactive

sdata = sd.read_zarr('data.zarr')

sdata['table'].obs['region'] = 'cell_labels'
sdata.set_table_annotates_spatialelement(
    table_name='table', 
    region='cell_labels', 
    instance_key='cell_labels'
)

Interactive(sdata)