DESeq2 with HTSeq DESeqDataSetFromHTSeqCount
1
0
Entering edit mode
@vincentpailler-16742
Last seen 5.7 years ago

Hi guys,

I'm trying to run DESeq2 on my HTSeq-count files

I've got 6 samples : 3 conditions with 2 replicates . I'd like to compare each condition between each other : 0h vs 6h , 0h vs 24h and 6h vs 24h

So my, my sampleTable looks like this :

> sampleTable
     samples  samples.1 condition
1 SRR2033948 SRR2033948        0h
2 SRR2033949 SRR2033949        0h
3 SRR2033950 SRR2033950        6h
4 SRR2033951 SRR2033951        6h
5 SRR2033952 SRR2033952       24h
6 SRR2033953 SRR2033953       24h

 

My dds looks like this :

> dds
class: DESeqDataSet
dim: 200542 6
metadata(1): version
assays(4): counts mu H cooks
rownames(200542): AT1G01010:exon:1 AT1G01010:exon:2 ...
  ATMG09960:exon:2 ATMG09980:exon:1
rowData names(26): baseMean baseVar ... deviance maxCooks
colnames(6): SRR2033948 SRR2033949 ... SRR2033952 SRR2033953
colData names(2): condition sizeFactor

 

I've created my 3 contrast comparisons :

res.A.B <- results(dds, contrast=c("condition","0h","6h"))

res.B.C <- results(dds, contrast=c("condition","6h","24h"))

res.A.C <- results(dds, contrast=c("condition","0h","24h"))

 

It works, but when I run resultsNames(dds) , I get

 

> resultsNames(dds)
[1] "Intercept"           "condition_24h_vs_0h" "condition_6h_vs_0h"

 

So, the contrast "condition_24h_vs_6h" doesn't appear and consequently, I can't shrink it.

> res_BC_LFC<-lfcShrink(dds,coef="condition_24h_vs_6h",type="apeglm") 
Error in lfcShrink(dds, coef = "condition_24h_vs_6h", type = "apeglm") :
  coef %in% resultsNames(dds) is not TRUE

 

So , I don't understand why I could run res.B.C <- results(dds, contrast=c("condition","6h","24h"))

 

Any suggestion why the contrast "condition_24h_vs_6h" doesn't appear?

 

Best,

Vincent

 

 

 

 

 

deseq2 • 514 views
ADD COMMENT
0
Entering edit mode
@mikelove
Last seen 1 day ago
United States

The contrast doesn’t appear because one can only fit the two comparisons to the reference, and additional coefficients would lead to linear dependence.

In the vignette we have some simple code for running nbinomWaldTest one more time to generate the LFC and SE needed for apeglm. Can you try this code and come back if there are additional problems or questions?

ADD COMMENT

Login before adding your answer.

Traffic: 949 users visited in the last hour
Help About
FAQ
Access RSS
API
Stats

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.

Powered by the version 2.3.6