Problem with neighbourhood identification and spatial analysis across TMAs/fovs (regardless of their image id)

[rach-crc]

Thank you for developing Scimap, it’s a really useful tool. However, I came across some questions when dealing with my data. And I wonder if you could kindly give me some advice. I have created a combined anndata ‘adata’ containing information about cells from different fovs (e.g. fov1/2/3/4/5/… or you can understand them as very small TMAs). Some of the fovs might come from same patients (e.g. fov 1-100 were from patient 1, fov 101-201 were from patient 2) and cell label id of each fov start from 1 (meaning that the coordinates for cells in this combined adata might overlap, as the size of all the fovs were same and we segmented cells from each fov individually).

In the tutorial page of Scimap, I found that most examples of spatial and neighbourhood analysis functions used adata of a single large image with merged/stitched fovs, and each cell has different coordinates. Hence, I wonder if there are ways to identify neighbourhoods across all the fovs regardless of their fov id and image id, meaning that we want to identify “common neighbourhoods” across all fovs. (e.g. cluster 1/2/3/4/5/… identified based on cells and their coordinates in each fov).

Meanwhile, I also have patient id and a particular gene mutation status for each cell, and I wonder if there are ways to use scimap functions for “across-all-fovs” neighbourhood identification and spatial analysis based on patient id and gene mutation status regardless of fov id and image id. (e.g., patient 1 identified neighbourhood cluster a b c based on all the fovs from that patient, patient 2 identified neighbourhood a c d e based one all the fovs from that patient, patient with gene mutation identified neighbourhood a b c and patient withou gene mutation identified neighbourhood a d e etc.)

The question for distance analysis function apply the same logic then, for example, comparing tumour-immune proximity in patients with mutation and those without.

Thank you for reading my message. I know it’s long, but I hope you can get the idea, and I would be very appreciated if you could give me some guidance.