As discussed during the VIB Ghent 2024 hackathon with @giovp and @melonora, removing the legacy _models.py from the codebase, and access the data directly form layer.metadata['sdata'] would improve the robustness of the code.
In particular we should rely on get_values() for fetching annotations instead of the currently used get_obs(), get_var() etc.
A positive implication of this refactoring is that the mismatch between the annotation indices shown by napari, and the indices used by SpatialData, would disappear, therefore fixing this bug: https://github.com/scverse/napari-spatialdata/issues/242.
As discussed during the VIB Ghent 2024 hackathon with @giovp and @melonora, removing the legacy
_models.py
from the codebase, and access the data directly formlayer.metadata['sdata']
would improve the robustness of the code.In particular we should rely on
get_values()
for fetching annotations instead of the currently usedget_obs()
,get_var()
etc.A positive implication of this refactoring is that the mismatch between the annotation indices shown by napari, and the indices used by
SpatialData
, would disappear, therefore fixing this bug: https://github.com/scverse/napari-spatialdata/issues/242.