Hi,
I've a small question concerning the differential expression analysis in DESeq2. So I've a 16 samples divided into 4 groups (control, ko1, ko2 and ko3 ). so one condition with 4 levels
as i understood form the vignette and previous questions regarding this matter here that performing wald statistics and separata pairwise comparisons would be the best options.
i am using a code like this one:
dds1 <- DESeqDataSetFromHTSeqCount(sampleTable=samples,directory="./",design= ~ condition) dds1 <- DESeq(dds1) res <- results(dds1)
Then pairwise comparison res.ko1 <- results(dds1, contrast=c("condition","ko1","ctrl")) res.ko2 <- results(dds1, contrast=c("condition","ko2","ctrl")) res.ko3 <- results(dds1, contrast=c("condition","ko3","ctrl"))
Now my question is, is it possible in DESeq to get a compined file of the significant DEG in the three cell types containing mainly how is the fold change is. somthing like a list with significant genes and the fold change in each level/ko.
Many thanks in advance
Thanks a lot for your feedback and answer. What about normalizing against the control. I couldn't find a clear approach on how to do that. As example i would like to normalize each genotype against the common control then compare them against each other. Is it possible to do that? Many thanks in advance.
This is asked often, comparing against a control is equivalent to comparing directly unless you have matched control samples. This follows from simple algebra:
(C - A) - (B - A) = (C - B)
Thanks a lot for your feedback. But what would be the case if i have a matched control specific to each experiment? I tried doing some kinda of a mixed condition as ( ~ condition + genotpye) but it didn't work out with me. Do you have some suggestion on how to tackle this if one needs to normalize against a specific control (as having a specific genotype for each control). So basically doing ((D-C) - (B-A)) ? Thanks in advance.
We have an example of paired data in the vignette FAQ, and of matches samples across groups
We have an example of paired data in the vignette FAQ, and of matched samples across groups in the section on interactions