Missing adata.h5ad in the explorer

[RitaRuoxinLi]

Hi,

I am using sopa to run baysor on our xenium samples. However, I noticed the the sopa (I reinstalled recently) doesn't give the output of h5ad in the <sample_folder>.explorer/ folder like before. Is this behavior expected? How should I get the adata (cell by gene matrix and the obs.slots) ?

I am appreciative of any suggestions.

This figure shows a sopa run which shows it is successful but no adata.h5ad output (right hand side shows the output file in the explorer)

This is a previous run 6 days ago.

[quentinblampey]

Hi @RitaRuoxinLi, thanks for reporting, this is unexpected.

Could you please show me the Snakemake config you use? And which version of Sopa do you use (2.0.0 or 2.0.1)?

Meanwhile, you can find your AnnData table inside your SpatialData object, i.e. the zarr directory which contains all the information:

import spatialdata

sdata = spatialdata.read_zarr("/path/to/data.zarr")

print(sdata) # see what's inside

adata = sdata["table"]
adata.write_h5ad("/path/to/data.h5ad")

[RitaRuoxinLi]

Hi Dr.Blampey,

Thank you for your response! The sopa version information are as follows:

1. sopa version is 2.0.1

2. Snakemake command
   snakemake --configfile=/home/workspace/sopa/workflow/config/xenium/baysor_multimodal_v1.yml --config data_path=/home/workspace/output-XETG00123__0036935__TIS08722-001__20240808__231722 --cores 30 --workflow-profile /home/workspace/sopa/workflow/profile/local

3. Configs

read:
  technology: xenium

patchify:
  patch_width_microns: 1750
  patch_overlap_microns: 20

segmentation:
  baysor:
    min_area: 20
    prior_shapes_key: cell_id # this allows to use the prior segmentation performed by 10X Genomics
    unassigned_value: UNASSIGNED

    config:
      data:
        force_2d: True # if false, uses 3D mode
        min_molecules_per_cell: 10
        x: "x"
        y: "y"
        z: "z"
        gene: "feature_name"
        min_molecules_per_gene: 0
        min_molecules_per_segment: 3
        confidence_nn_id: 6

      segmentation:
        scale: 6.25 # typical cell radius
        scale_std: "25%" # cell radius standard deviation
        prior_segmentation_confidence: 0.5
        estimate_scale_from_centers: false
        n_clusters: 4
        iters: 500
        n_cells_init: 0
        nuclei_genes: ""
        cyto_genes: ""

aggregate:
  average_intensities: true
  min_transcripts: 10 # [optional] cells whose transcript count is below that this threshold are filtered

[quentinblampey]

I found the issue: it appeared in sopa==2.0.1 when adding a table_key argument in the CLI for the explorer.

It should be fixed now, and it'll be released in sopa==2.0.2!

Meanwhile, you can also use the master branch if you want to test this feature, or run (on your current installation) the CLI with the right table_key argument:

sopa explorer write /home/workspace/output-XETG00123__0036935__TIS08722-001__20240808__231722.zarr --mode '-i' --table-key table

[quentinblampey]

I just released sopa==2.0.2, it should be now fixed!