match_sdata_to_table removes layers

[MeyerBender]

Using the method match_sdata_to_table() removes a lot of layers, which is undesirable.

Minimal example:

import spatialdata as sd
from spatialdata.datasets import blobs

sdata = blobs()
sdata

This returns:

SpatialData object
├── Images
│     ├── 'blobs_image': DataArray[cyx] (3, 512, 512)
│     └── 'blobs_multiscale_image': DataTree[cyx] (3, 512, 512), (3, 256, 256), (3, 128, 128)
├── Labels
│     ├── 'blobs_labels': DataArray[yx] (512, 512)
│     └── 'blobs_multiscale_labels': DataTree[yx] (512, 512), (256, 256), (128, 128)
├── Points
│     └── 'blobs_points': DataFrame with shape: (<Delayed>, 4) (2D points)
├── Shapes
│     ├── 'blobs_circles': GeoDataFrame shape: (5, 2) (2D shapes)
│     ├── 'blobs_multipolygons': GeoDataFrame shape: (2, 1) (2D shapes)
│     └── 'blobs_polygons': GeoDataFrame shape: (5, 1) (2D shapes)
└── Tables
      └── 'table': AnnData (26, 3)
with coordinate systems:
    ▸ 'global', with elements:
        blobs_image (Images), blobs_multiscale_image (Images), blobs_labels (Labels), blobs_multiscale_labels (Labels), blobs_points (Points), blobs_circles (Shapes), blobs_multipolygons (Shapes), blobs_polygons (Shapes)

Then, if one runs:

sdata = sd.match_sdata_to_table(sdata, "table")
sdata

The resulting object looks like this:

SpatialData object
├── Labels
│     └── 'blobs_labels': DataArray[yx] (512, 512)
└── Tables
      └── 'table': AnnData (26, 3)
with coordinate systems:
    ▸ 'global', with elements:
        blobs_labels (Labels)

Desired behavior: keep all layers, and subset them according to the cells present in the table. For layers that cannot be filtered (such as the image), I would like to keep them in the object as is.

[LucaMarconato]

Thanks for reporting. This is a limitation of the design and not a consequence of mast_sdata_to_table(): the table can only map to a single element.

Still, it's worth to take this feedback into account and think if a larger modification of the structure could help.

[MeyerBender]

I think this would indeed be super helpful. One of the main selling points of scanpy is the smooth and synchronized subsetting. Having something like this for spatialdata would also make things such as spatial queries much more intuitive and useful.