add the option to plot filtered out cells to main

[lopollar]

in the branch add_filtered _cells, some code was added to plot the cells that were filtered out in red to the plots. This looks as follows: As this is very useful for quality control, is should be added into the new functions.

@ArneDefauw, do you have time to look at this, or should I look into it?

It comes down to saving the cells that were filtered in different shapes layers and then plotting all the filtered layers.

[ArneDefauw]

This is still supported on main. The filtered out shapes are added to the sdata object:

├── Shapes
│     ├── 'expanded_cells20': GeoDataFrame shape: (1463, 1) (2D shapes)
│     ├── 'filtered_expanded_cells20_low_counts': GeoDataFrame shape: (52, 1) (2D shapes)
│     ├── 'filtered_expanded_cells20_segmentation': GeoDataFrame shape: (7, 1) (2D shapes)
│     ├── 'filtered_expanded_cells20_size': GeoDataFrame shape: (33, 1) (2D shapes)
│     ├── 'filtered_segmentation_mask_boundaries_low_counts': GeoDataFrame shape: (52, 1) (2D shapes)
│     ├── 'filtered_segmentation_mask_boundaries_segmentation': GeoDataFrame shape: (7, 1) (2D shapes)
│     ├── 'filtered_segmentation_mask_boundaries_size': GeoDataFrame shape: (33, 1) (2D shapes)
│     └── 'segmentation_mask_boundaries': GeoDataFrame shape: (1463, 1) (2D shapes)

and the filtered out cells can be plotted via

nas.pl.plot_shapes( sdata, img_layer='clahe', shapes_layer='segmentation_mask_boundaries', plot_filtered=True )

resulting in:

While checking this feature I found a small bug. If more than one shapes layer is added (e.g. if you did cell expansion with voronoi), then setting plot_filtered=True would plot all filtered cells, also the voronoi shapes. Therefore I changed to code in _plot_shapes from

                if f"filtered" in i:
                    sdata[i].cx[crd[0] : crd[1], crd[2] : crd[3]].plot(
                        ax=ax,
                        edgecolor="red",
                        linewidth=1,
                        alpha=alpha,
                        legend=True,
                        aspect=1,
                        cmap="gray",
                    )

to

                if f"filtered_{shapes_layer}" in i:
                    sdata[i].cx[crd[0] : crd[1], crd[2] : crd[3]].plot(
                        ax=ax,
                        edgecolor="red",
                        linewidth=1,
                        alpha=alpha,
                        legend=True,
                        aspect=1,
                        cmap="gray",
                    )

I will push this fix on main later today

[ArneDefauw]

this is supported on main since https://github.com/saeyslab/napari-sparrow/pull/142