DESEQ PCA
1
0
Entering edit mode
PJC • 0
@5b1ac7f9
Last seen 28 days ago

Hi

I'm having some real problems with a miRNA analysis.

I have 36 samples from 4 groups. The four groups share features (9 YA, 9 YB, 9XA, 9XB).

I want to look at differential expression between Y vs X, A vs B, YA vs YB, & XA vs XB

I'm using deseq2 for differential analysis:

dds <- DESeqDataSetFromMatrix(
colData = coldata,
design= ~0 + condition)


To obtain my results I'm using the following

res_YX<-results(dds, contrast = list(c("conditionYA", "conditionYB"),
c("conditionXA", "conditionXB")), listValues = c(1,-1), alpha=0.05)

res_AB<-results(dds, contrast = list(c("conditionYA", "conditionXA"),
c("conditionYB", "conditionXB")), listValues = c(1,-1), alpha=0.05)

res_YAYB<-results(dds, contrast = list(c("conditionYA"),
c("conditionYB")), listValues = c(1,-1), alpha=0.05)

res_XAXB<-results(dds, contrast = list(c("conditionXA"),
c("conditionXB")), listValues = c(1,-1), alpha=0.05)


When I put together a PCA plot there really isn't any grouping according to the condition, even I look at the larger groups (Y vs X, A vs B). I have looked at other principle components and excluded potential outliers and it doesn't really help.

Are there any good resources to troubleshoot this or advice about how to proceed?

Thanks

PJC

DESeq2 miRNA • 172 views
0
Entering edit mode

The data is what it is. You can't 'fix' it just because you don't like what the PCA shows you. You might still have useful DE genes even if the first few PCs don't look the way you'd wish.

0
Entering edit mode

See some of the comments here about designs: DESeq2 comparison two cell types and two genotypes

0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

Agree with the above comment. This is what the PCA is for, it’s an exploratory plot that you don’t need to adjust.