I'm trying to perform differential gene expression analyses using DESeq, on a pre-existing dataset, where 8 cell types (A to H) were identified from MARS-Seq on whole organism. Gene expression counts are given for
A x 9 replicates
B x 1
C x 1
D x 3
E x 1
F x 1
G x 4
H x 1
In order to find genes differentially expressed in A, I contrast A against "notA" (replacing B to H with "notA")
dds<-DESeqDataSetFromMatrix(countData = cts, colData = coldata,design=~condition) dds$condition<-factor(dds$condition,levels=c("A","notA")) dds<-DESeq(dds) alpha<-0.1 res<-results(dds,contrast=c("condition","A","notA"),alpha=alpha) reslfc<-lfcShrink(dds,contrast=c("condition","A","notA"),res=res, alpha=alpha)
and so on for the other 7 cell types.
However, I found that I get much more DEGs for A D & G, i.e. samples with more reps. In fact, I have no DEGs at all for B, C E, F and H.
Anyway I can overcome this? Or should I not be performing the analysis this way?