Inconsistent FoldChanges in FindMarkers for different tests

[morallawwithin]

I noticed that there are differences in FC when using different types of tests for DE. The magnitude of these changes is somehow related to the number of feature that each test reports back. I could reproduce this bug with the ifnb dataset, but on my own data these difference are even higher (sometimes up to 1).

library(Seurat)
library(SeuratData)
library(DESeq2)
InstallData("ifnb")
ifnb <- LoadData("ifnb")
ifnb <- NormalizeData(ifnb)
ifnb$celltype.stim <- paste(ifnb$seurat_annotations, ifnb$stim, sep = "_")
Idents(ifnb) <- "celltype.stim"
mono.Wilcox <- FindMarkers(ifnb, ident.1 = "CD14 Mono_STIM", ident.2 = "CD14 Mono_CTRL")
mono.DESeq <- FindMarkers(ifnb, ident.1 = "CD14 Mono_STIM", ident.2 = "CD14 Mono_CTRL", test.use = "DESeq2")

Notice the difference between

head(mono.DESeq, n = 3)
       p_val avg_log2FC pct.1 pct.2 p_val_adj
CCL8       0   9.225442 0.918 0.017         0
CXCL11     0   8.789423 0.932 0.012         0
CXCL10     0   8.335510 0.984 0.035         0

and

head(mono.Wilcox, n = 3)
       p_val avg_log2FC pct.1 pct.2 p_val_adj
CCL8       0   9.134191 0.918 0.017         0
CXCL11     0   8.624208 0.932 0.012         0
CXCL10     0   8.036564 0.984 0.035         0

while

nrow(mono.DESeq)
[1] 7347
nrow(mono.Wilcox)
[1] 6956

From the documentation it is clear that test.use only influences the p_val and the features that get returnd.

[mhkowalski]

Hi,

Thanks for reaching out and for you providing a reproducible example. What version of Seurat are you using? When I run with most recent versions, I don't see a discrepancy.

> head(mono.Wilcox)
        p_val avg_log2FC pct.1 pct.2 p_val_adj
IFIT1       0   7.319139 0.985 0.033         0
CXCL10      0   8.036564 0.984 0.035         0
RSAD2       0   6.741673 0.988 0.045         0
TNFSF10     0   6.991279 0.989 0.047         0
IFIT3       0   6.883785 0.992 0.056         0
IFIT2       0   7.179929 0.961 0.039         0
> head(mono.DESeq)
        p_val avg_log2FC pct.1 pct.2 p_val_adj
IFIT1       0   7.319139 0.985 0.033         0
CXCL10      0   8.036564 0.984 0.035         0
RSAD2       0   6.741673 0.988 0.045         0
TNFSF10     0   6.991279 0.989 0.047         0
IFIT3       0   6.883785 0.992 0.056         0
IFIT2       0   7.179929 0.961 0.039         0