Q: How to render xenium morphology image with original colors?

[tsvvas]

Hello @LucaMarconato!

Thank you very much for this tool. I want to plot a part of morphology image acquired with xenium + multimodal segmentation kit. It is loaded into spatialdata and plotted without errors, but the colors are dim and they are different from what I'd expected (DAPI - #0F73E6, Boundary - #F300A5, Interior RNA - #A4A400, interior protein - #008A00, reference values from Xenium Explorer). I tried passing a list of colors like this:

crop = lambda sdata: spatialdata.bounding_box_query(
    sdata,
    min_coordinate=[17_500, 55_000],
    max_coordinate=[19_500, 57_000],
    axes=("x", "y"),
    target_coordinate_system="global",
)

crop(sdata).pl.render_images("morphology_focus", ["#0F73E6", "#F300A5", "#A4A400", "#008A00"]).pl.show(
    ax=axes[0], title="Morphology image", coordinate_systems="global"
)

The code does not produce an error, it works twice as long, and it results in the same image with colors unchanged.

The key question is how to adjust the colors and saturation of the plot, so that it looked closer to what I see in Xenium Explorer?

Secondary question is how to translate physical coordinates into spatialdata's global coordinate system? Am I supposed to get transformation and then apply transformation to a single or a couple points to get min and max for cropping above? I can make a separate issue with this question, if necessary.

Best,
Vasily

[LucaMarconato]

Hi, for coloring the channels please use the palette argument.

Regarding coordinates yes, once you identify the transformation from the physical coordinate system to the global one, you would need to transform the coordinates and use these in the bounding box query.

We are going to improve the ergonomics of this as described in these issues (except for 338, which is unrelated): https://github.com/scverse/spatialdata/issues?q=sort%3Aupdated-desc%20is%3Aissue%20is%3Aopen%20physical

[LucaMarconato]

Also please check my comment here (which answers almost a duplicate of question 2): scverse/spatialdata-io#205

[timtreis]

Hey @tsvvas,

I assume you have an image with 4 channels that you're trying to map the individual colors to, right? As Luca said, you can use the palette argument for this, see here:

Please let us know if that solves your issue :)