correct 2um bins positions and counts

[alikhuseynov]

Hi, thanks for this very useful tool!
Just few comments/thoughts:

• Do we see non-uniform spot distribution in VisiumHD? Similar to old 100um array printing artifacts (not sure if this was fixed for Visium at 55um spots), see this paper -> Figure S1. Such positional differences can be corrected using Centroidal Voronoi tessellation. I wonder if this will alleviate the typical "striped" pattern of bins.
• Alternative for UMI counts correction (destripe). Using instead spatial continuity-based neighborhood graph (queen or rook pattern) then aggregating or doing other stats within each neighborhood to correct observed spot count.

[nadavyayon]

Hey, Thanks for these suggestions.

As far as I am aware, visiumHD doesn't have positional warping.

As to the destriping, the effects are clearly row and column dependent, so intuitively, I don't think a local correction would be a good idea.

N