You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
in both nanostring cosmx and xenium we available info on labels and labels' centroids. Therefore, the spatialdata object will (could) have 2 regions representing the same instance (with potentially same instance_id) and a single table annotating both.
I believe we discussed this but I don't remember how we ended up resolving this issue.
I think one option we discussed was that the centroids labels would just be stored in table.obsm["spatial"] essentially as an annotation on the labels (and useful for plotting etc) but without any constraint or specifications (e.g. no coordinate systems and/or transforms). If that's the case, then there would be no Shapes region for the centroids.
Hi, from the discussion we settled, as you mentioned, on storing the centroids in the table as a separate column, and if I recall correctly we said to store the centroids in the implicit coordinates. This is meant to be a way to unblock this now and not the definitive solutions, since if for instance the centroids come from a labels and they are cropped, then the implicit coordinates stored in the table would not be correct anymore. I also agree of removing the Shapes from storage, they could be created in-place at any time if needed.
Activity
LucaMarconato commentedon Jan 8, 2023
Hi, from the discussion we settled, as you mentioned, on storing the centroids in the table as a separate column, and if I recall correctly we said to store the centroids in the implicit coordinates. This is meant to be a way to unblock this now and not the definitive solutions, since if for instance the centroids come from a labels and they are cropped, then the implicit coordinates stored in the table would not be correct anymore. I also agree of removing the
Shapesfrom storage, they could be created in-place at any time if needed.kevinyamauchi commentedon Jan 9, 2023
My recollection is the same as @LucaMarconato 's.
LucaMarconato commentedon Feb 22, 2024
Closed as duplicate of #34 and #123.