Hi,
I would like to perform multiple conditions analysis in one go. I can see in EdgeR that it is possible to do like glmLRT(fit, coef=2:4). What is the equivalent way of doing this in DESeq2?
Does design=time + daylength + salinity do the job?
Thank you for the help.
J.
See the vignette section on likelihood ratio tests in DESeq2, and then you can read the LRT section of ?DESeq.
Hi Michael,
I have read the documents did the following.
ddsHTSeq<-DESeqDataSetFromHTSeqCount(sampleTable=sampleTable, directory=INDIR, design=~time + daylength + salinity)
dds <- DESeq(ddsHTSeq,betaPrior=FALSE, test="LRT", full=~time + daylength + salinity, reduced=~daylength)
Now I have problem when I have only one level for two conditions and three levels for one conditions as you can see below. I have the following:
time=all samples T4
daylength=LD,SP,SPLD
salinity= all samples FW
My samples are (6 replications each):
T4-LD-FW[1-6]
T4-SP-FW[1-6]
T4-SPLDFW[1-6]
I want to do statistical test on daylength which has three factors. Is it possible to do statistical test for all three (LD vs SP vs SPLD) together?
Thanks a lot Michael.
hi John, a minor note about the support site: you can use the blue ADD REPLY and ADD COMMENTS buttons if you want to reply to an existing answer.
If you were answering someone else's post you would use the green Add Answer button.
It's a bit easier to follow the discussion if you use the comment/reply buttons because this lead to threaded discussions, where one can see what answer you are replying to (the answers will get reordered if someone clicks the thumbs up button and then it would be hard to follow the order of the discussion).
Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
version 2.3.6
hi John,
For this experiment, note that you have three groups of samples: those in LP, SP and SPLD. As there are no differences in T4 and FW across these samples, you need not add these to the design (and indeed it will produce an error, because these variables will be collinear with the intercept).
Just full=~ daylength vs reduced=~ 1