calcStateChanges error when contamination = T

[marsdenl]

Hey,

Thanks for developing this tool :) I ran through the workflow using some merscope data and whilst everything ran fine without correction for contamination, I ran into an issue of I do correct for it. Here is the workflow and error:

merscope <- getDistances(merscope,
imageID = 'sample',
spatialCoords = c("sdimx", "sdimy"),
cellType = "cluster",
maxDist = 200,
nCores = 1
)

merscope <- getAbundances(merscope,
imageID = 'sample',
spatialCoords = c("sdimx", "sdimy"),
cellType = "cluster",
r = 200,
nCores = 10
)

merscope <- calcContamination(merscope, cellType = "cluster", assay = 'X') #X is lognormalized counts

contam <- reducedDim(merscope, "contaminations")
reducedDim(merscope, "contaminations") <- contam[, !colnames(contam) %in% c("cluster")]

stateChanges_corr <- calcStateChanges(
cells = merscope,
type = "distances",
imageID = 'sample',
cellType = "cluster",
nCores = 1,
minCells = 100,
contamination = TRUE
)

Error: BiocParallel errors
1 remote errors, element index: 7
143 unevaluated and other errors
first remote error:
Error in [.data.frame(design, , nzv, drop = FALSE): undefined columns selected

If contamination = False, it runs fine.

Do you know what might have happened?

Thanks :)