Missing filtered table in polygon_query() of MERFISH data

Using this tutorial for MERFISH data, I found that the output of polygon_query(sdata, my_napari_annotation) does not contain any filtered table, despite my_napari_annotation covering multiple cells. Is there something missing in my code?

With filter_table=False, I get the unfiltered table, which is not what I am looking for. This is similar to https://github.com/scverse/spatialdata/issues/532

Much appreciated!

Details:

# load

sdata = sio.merscope(full_paths, backend="rioxarray", slide_name="s1", region_name="r0")

sdata
SpatialData object
├── Points
│     └── 's1_r0_transcripts': DataFrame with shape: (<Delayed>, 9) (2D points)
├── Shapes
│     └── 's1_r0_polygons': GeoDataFrame shape: (96632, 9) (2D shapes)
└── Tables
      └── 'table': AnnData (96632, 500)
with coordinate systems:
    ▸ 'global', with elements:
        s1_r0_transcripts (Points), s1_r0_polygons (Shapes)

sdata.table.uns
OrderedDict([('spatialdata_attrs',
              {'region': 's1_r0_polygons',
               'region_key': 'cells_region',
               'instance_key': 'EntityID'})])

sdata["table"]
AnnData object with n_obs × n_vars = 96632 × 500
    obs: 'fov', 'volume', 'center_x', 'center_y', 'min_x', 'min_y', 'max_x', 'max_y', 'anisotropy', 'transcript_count', 'perimeter_area_ratio', 'solidity', 'DAPI_raw', 'DAPI_high_pass', 'PolyT_raw', 'PolyT_high_pass', 'region', 'slide', 'dataset_id', 'cells_region', 'EntityID'
    uns: 'spatialdata_attrs'
    obsm: 'blank', 'spatial'

print(sd.SpatialData.get_annotated_regions(sdata["table"]))
s1_r0_polygons

# annotate shape

interactive = napari_spatialdata.Interactive(sdata)

interactive.run() 

# export shape layer from Napari

shape = "overlap"

sdata[shape]
geometry
0	POLYGON ((5423.45 1777.948, 6398.625 1777.948,...

sdata[shape].geometry.iloc[0]

sdata
SpatialData object
├── Points
│     └── 's1_r0_transcripts': DataFrame with shape: (<Delayed>, 9) (2D points)
├── Shapes
│     ├── 'overlap': GeoDataFrame shape: (1, 1) (2D shapes)
│     └── 's1_r0_polygons': GeoDataFrame shape: (96632, 9) (2D shapes)
└── Tables
      └── 'table': AnnData (96632, 500)
with coordinate systems:
    ▸ 'global', with elements:
        s1_r0_transcripts (Points), overlap (Shapes), s1_r0_polygons (Shapes)

# sd.polygon_query

subset = sd.polygon_query(
    sdata, polygon=sdata[shape].geometry.iloc[0], target_coordinate_system="global", filter_table=True
)

subset
SpatialData object
└── Shapes
      └── 'overlap': GeoDataFrame shape: (1, 1) (2D shapes)
with coordinate systems:
    ▸ 'global', with elements:
        overlap (Shapes)

# subset does not contain filtered table

Reproduce the documented MERFISH workflow and the shown polygon_query() call with filter_table=True, then compare it with filter_table=False and review the related issue 532. Done means the resulting SpatialData object includes the filtered table for the annotated polygon.