Hello,

I have three groups for RNAseq: 1."control" (nothing), 2."sw"treatment and 3."sw.bp" bacteria + treatment (unfortunately the bacteria alone cannot be cultured). I would like to know the differentially expressed genes triggered by the bacteria. I created the following contrast matrix, but all the genes are differentially expressed...I was wondering how should i create a proper contrast matrix as probably I'm having an issue here. Thanks!

> 
> SWvsSWBPCNT <- makeContrasts(sw.bp-sw-control, levels = design)
> 
> DESWvsSWBPCNT <-glmQLFTest(fit, contrast=SWvsSWBPCNT)
> topTags(DESWvsSWBPCNT)


Coefficient:  -1*control -1*sw 1*sw.bp 
                                genes   SYMBOL ENTREZID logFC logCPM     F   PValue      FDR
ENSMUSG00000013698 ENSMUSG00000013698   Pea15a    18611  14.1   6.00 18995 1.98e-20 4.71e-18
ENSMUSG00000040818 ENSMUSG00000040818  Dennd6a   211922  14.2   5.85 17841 2.87e-20 4.71e-18
ENSMUSG00000028851 ENSMUSG00000028851     Nudc    18221  14.0   6.14 17657 3.05e-20 4.71e-18
ENSMUSG00000057554 ENSMUSG00000057554   Lgals8    56048  13.6   6.26 17315 3.43e-20 4.71e-18
ENSMUSG00000028991 ENSMUSG00000028991     Mtor    56717  13.8   6.03 17272 3.48e-20 4.71e-18
ENSMUSG00000099083 ENSMUSG00000099083     Atf7   223922  14.2   5.83 17205 3.56e-20 4.71e-18
ENSMUSG00000028223 ENSMUSG00000028223    Decr1    67460  14.2   5.72 17094 3.70e-20 4.71e-18
ENSMUSG00000023932 ENSMUSG00000023932    Cdc5l    71702  13.8   6.10 17048 3.76e-20 4.71e-18
ENSMUSG00000023367 ENSMUSG00000023367 Tmem176a    66058  13.5   6.39 16905 3.95e-20 4.71e-18
ENSMUSG00000026280 ENSMUSG00000026280    Atg4b    66615  14.1   5.74 16884 3.98e-20 4.71e-18

> is.de <- decideTestsDGE(DESWvsSWBPCNT)
> summaryis.de)
       -1*control -1*sw 1*sw.bp
Down                          0
NotSig                        1
Up                        15297

No, that's not sensible. You possibly intended:

makeContrasts(sw.bp-(sw+control)/2, levels = design)

Alternatively you could compare sw.bp to sw and control separately, perhaps combining with an F-test.