ArchR multiome pipeline getMarkerFeatures()： pbmc_10k demo data failed with import groups

[Jiahuan90]

Hi rcorces,

I ran ArchR multiome pipeline with pbmc_10k demo data.

I encounter a problem, I use Seurat to determine the cell type and import the cell type to ArchR project. I want to use getMarkerFeatures() to find out the pbmc_10k specific peaks of each cell type. However, it failed with "Found less than 100 cells for background matching, Lowering k to 0: Error in nabor::knn(data = data, query = query, k = k, ...) : Cloud has no points".

However, I am sure each cell type has >100 cells but it get the error.

My code is:
markersPeaks <- getMarkerFeatures(
ArchRProj = proj,
useMatrix = "PeakMatrix",
groupBy = "cell.type",
bias = c("TSSEnrichment", "log10(nFrags)"),
testMethod = "wilcoxon",
binarize = TRUE
)

Meanwhile, I could use getMarkerFeature() to distinguish specific peaks on group.by = "Clusters" (LSI combined) ----- successful

I am not sure if cell.type has NA affect the getMarkerFeatures? since I have checked ArchR project$cell.type, and found 87 cells marked NA? -------------- I have tried, removed markded NA cells, and run getMarkerFeatures() with groupBy = "cell.type" failed.

Is the import cell.type information also need to do LSI reduction or something else?

Thank you for your help~