DESeq can not detect significant DE genes.

[lzlgboy]

Thank you very much blancha~
Works like a charm.

[blancha]

You just reversed the order of the treatment and control conditions.
In your counts file, it's Treat1, Treat2, Treat3, CTR1, CTR2,
In your conditions vector, it's CTR, CTR, Treat, Treat, Treat.
It's actually reassuring that DESeq does not find any significant differences in gene expression with this input.

> conditions.CTR.vs.Treat
[1] CTR CTR Treat Treat Treat
Levels: CTR Treat
> head(countTable.CTR.vs.Treat)
Treat1 Treat2 Treat3 CTR1 CTR2
Mbp 255219 261306 295123 365033 415838
Apoe 255618 333022 275713 31494 33771
Gfap 135100 143838 143669 57978 62635
Plp1 83787 92153 100662 165798 207643
Fth1 97740 108155 113955 122369 117290
Sparcl1 44654 43945 53549 72772 90997

[lzlgboy]

Dear all,
I recently used DEseq to call DE genes and I have successfully run it for several dataset. However for the following dataset, I could not get any 'significant' DE gene with Padj < 0.05. Though, the within group variation between replicates are small and from the correlation between samples you can tell that CTR and Treated are obviously different. So my question is how could this happened? I have tried to figure out the problem but with no success.

Here is my R command for the analysis:
input file are attached behind. (ExperimentDesign.txt and RawCountTable.txt)
=============================
countTable.TBI.all = read.table("RawReadCount.txt",header=TRUE, row.names=1)
designTable.TBI.all = read.table("ExperimentDesign.txt",header=TRUE, row.names=1)
samples.CTR.vs.Treat = designTable.TBI.all$Condition == "CTR" | designTable.TBI.all$Condition =="Treat"
countTable.CTR.vs.Treat = countTable.TBI.all[ , samples.CTR.vs.Treat]
conditions.CTR.vs.Treat = designTable.TBI.all$Condition[samples.CTR.vs.Treat]
conditions.CTR.vs.Treat = factor(conditions.CTR.vs.Treat)
library(DESeq)
CTR.vs.Treat.cds = newCountDataSet(countTable.CTR.vs.Treat, conditions.CTR.vs.Treat)
CTR.vs.Treat.cds = estimateSizeFactors(CTR.vs.Treat.cds)
CTR.vs.Treat.cds = estimateDispersions(CTR.vs.Treat.cds)
CTR.vs.Treat.res = nbinomTest(CTR.vs.Treat.cds, "CTR", "Treat")
===================================================