This is more a nice-to-have feature request, but I came across this now a couple of times. For me, it would be nice, if liana_wrap would inherit the levels of the cluster identities, because this would ensure consistent order of clusters in plotting. Using the tutorial example:
Here, the clusters in the testdata Seurat object are a factor:
> levels(Idents(testdata))
'B''CD8 T''NK'
and it would be nice if those levels would be inherited to source and target in liana_test, I think. Currently, liana_test does not return factors for those:
> levels(liana_test$source)
NULL
Of course, one can just convert the source and target columns to factors manually as a workaround.
This is more a nice-to-have feature request, but I came across this now a couple of times. For me, it would be nice, if
liana_wrap
would inherit the levels of the cluster identities, because this would ensure consistent order of clusters in plotting. Using the tutorial example:Here, the clusters in the
testdata
Seurat object are a factor:and it would be nice if those levels would be inherited to
source
andtarget
inliana_test
, I think. Currently,liana_test
does not return factors for those:Of course, one can just convert the source and target columns to factors manually as a workaround.