cbind: unique/duplicated colnamesare left asis/made unique; read10xVisium: keep barcodes as colData; spatialCoords/<-: withDimnames argument

[HelenaLC]

Addressing #100 ...

• cbind assures unique/duplicated colnamesare left asis/made unique
• in case of duplicates, column names are prefixed with n_... where
  n corresponds to the number of the object being combined
• to retain original barcode identities, these are now kept as colData
  by read10xVisium (previously used as row names and dropped)

... and #103

• added argument withDimnames to spatialCoords getter & setter
• discrepancies b/w colnames(x) and rownames(spatialCoords(x)) throw an error,
  unless withDimnames=FALSE or the latter are NULL

[HelenaLC]

So, can someone accept this if we're okay with the GA (& changes, ofc)?

[drighelli]

the GHA fail during the tests:

── Error ('test_SpatialExperiment-spatialCoords.R:27'): spatialCoords<-,matrix ──
Error in `rownames(xyz) <- NULL`: object 'xyz' not found

do you know why?

[drighelli]

there are no xyz in the read10xVisium example, I suppose the GHA error comes from here...

[HelenaLC]

Damn it, I had reversed the order of some lines... should be fixed now.

[drighelli]

there is a warning that I suppose is coming from the "matrix" instead of a "value"

Warning:     ‘spatialCoords<-’
    ‘\S4method{spatialCoords<-}{SpatialExperiment,NULL}’
    ‘\S4method{spatialCoords<-}{SpatialExperiment,matrix}’
  The argument of a replacement function which corresponds to the right
  hand side must be named ‘value’.

[HelenaLC]

Yeah I saw. But I don’t understand because it’s good locally. I’m on it…

[drighelli]

indeed I don't get why, but have you tried something like: c("SpatialExperiment", "value") ?

[drighelli]

I think that's because the generic is defined as setGeneric("spatialCoords<-", function(x, value, withDimnames=TRUE) standardGeneric("spatialCoords<-"))

[HelenaLC]

…the signature has to be classes, so “NULL” and “matrix” (not “value”) is correct. It’s saying the generic and methods don’t match. But I don’t see why not as they both have “value” in the function definition in the same order…

[drighelli]

yes, I saw that, I'm looking over the internet, but still I'm not able to understand the motivation for this warning.