Hi,

Working on a scRNAseq data set from inflamed murine skin.

I have done cluster analysis using Seurat. Upon using the FindAllMarkers function and looking at the top 3 markers of each cluster, it is somewhat clear to me what cell type each cluster represents. I can tell what is a T-cell, monocyte, macrophage, stromal cell, etc...

However, when I used SingleR to get cell type annotations on the SCE object using various reference data, I find that the immune cell annotations are no longer specific to a cluster. I have tried different reference data.

Here is my UMAP plot without annotation assignments

and there is my UMAP plot with annotations. for this one, I used ImmGen as a reference. Annotations are completely dispersed.

How do I ensure cluster specific annotations?

Script:

   immgenref <- celldex::ImmGenData()

pac10sce <- as.SingleCellExperiment(DietSeurat(pac10))


immgen.main <- SingleR(test = pac10sce,assay.type.test = 1,ref = immgenref,labels = immgenref$label.main)
immgen.fine <- SingleR(test = pac10sce,assay.type.test = 1,ref = immgenref,labels = immgenref$label.fine)

table(immgen.main$pruned.labels)
table(immgen.fine$pruned.labels)

pac10@meta.data$immgen.main <- monaco.main$pruned.labels

pac10<- SetIdent(pac10, value = "immgen.main")
DimPlot(pac10, label = T , repel = T, label.size = 3)