Open multimeric opened 1 year ago
Makes sense, but an AnnData could, in principle, contain a layer named X
. I would add a check if that's the case and some way to handle the collision, e.g. naming the main assay XX
or main-X
or something like that (also checking for collisions with the modified name)
I take your point, I guess that could happen, although it would be a very weird and confusing AnnData file. I'm fairly sure that Zellkonverter just gives it the name "X", but any name is better than nothing.
It doesn't seem to be required that all the assays are named, but I've noticed it breaks various downstream code if it is not. Most notably
cbind(some_sce)
will fail if not all assays are named.