Analyze cohorts

[LukasHats]

Dear @ajitjohnson,

Is there a way in the spatial interaction analysis to pass a variable containing cohort information? Lets say I have a adata.obs called disease, would it somehow be possible to pass that to the plotting or calculation function to compare the different disease cohorts? I suspect that passing disease to the imageid variable in sm.tl.spatial_interaction is not correct right? If I am doing this, I will get a heatmap comparing the disease cohorts when summarize_plot is set to False but I guess this messes with the coordinates right? Would the only solution be to split the anndata based on the disease column and run it separately and then compare? Or do you have other suggestions?

Further, for the proximity scoring, would it be possible to set the score_by = 'disease?

Thanks in advance!