Seurat: Differential Expression Not Returning All Genes

[lothm2014]

Hello,

I am running differential expression using the FindMarkers function. I would like to return all the genes that are in the original counts matrix. My attempt to include all genes is the following:

A431DeResults<- FindMarkers(seuratObject2, ident.1 = 0, ident.2 = 2, logfc.threshold=0,  min.pct=0, 
min.diff.pct=0, thresh.test=0)

The result includes 16,171 genes, but there are ~26,000 genes in the counts matrix. How can I get results that include all of the genes?

[nicodemus88]

Hi @lothm2014,
Try changing some of the parameters as below:

degRes <- FindMarkers(object = so, ident.1 = "Tx", ident.2 = "Ctrl", min.pct = -Inf, logfc.threshold = -Inf, min.cells.feature = 1, min.cells.group = 1)

[xiaokang8002]

Hi @lothm2014,
Try changing some of the parameters as below:

degRes <- FindMarkers(object = so, ident.1 = "Tx", ident.2 = "Ctrl", min.pct = -Inf, logfc.threshold = -Inf, min.cells.feature = 1, min.cells.group = 1)

@nicodemus88
df <- FindMarkers(object, ident.1 = "cluster 12", ident.2 = "cluster 12", verbose = FALSE, logfc.threshold = -Inf,
min.pct = -Inf,min.cells.feature = 0,min.cells.group = 0)
I used the above code. the "object" had 23407 features. Why the "df" just had 2000 features?
Thanks a lot!

[nicodemus88]

Because the differential expression is performed on the original assay, while the df has only the number of variable features you used.

[xiaokang8002]

Because the differential expression is performed on the original assay, while the df has only the number of variable features you used.

Hi nicodemus88,
Thanks for your prompt comment.
TLH.list <- lapply(X = TLH.list, FUN = function(x) {
x <- NormalizeData(x)
x <- FindVariableFeatures(x, selection.method = "vst", nfeatures = 2000)
})
Was it due to "nfeatures = 2000" in the above code?
Or could you tell me how to let "df" had all the genes of "object"?
Thanks!

[nicodemus88]

@xiaokang8002,
Your data has all the genes (23,407 features), it's only that you have a subset of it as variable features, in this case 2000 features, which is used to perform dimensional reduction.
When plotting or performing differential expression tests, the original assay (the data slot in RNA assay) is used so all genes are included.