Paired-test in DESeq or DESeq2

[gcai]

Hi,

I have an experiment with 7 individuals. Each was sampled before and after treatment. It seems to me that I should do a paired-test. Anyone knows how to do that in DESeq or DESeq2?

[dpryan]

You just specify the individual as a blocking factor (e.g., ~individual+time).

[gcai]

Thank you Devon.

I have 7 individuals that were sampled before treatment (t0) and at t1, t2, t3, and t4 after treatment. To use the individual as blocking factor, and estimateDispersion (in DESeq):
cds = estimateDispersions(cds, method="pooled-CR")

I got the warning message:
glm.fit: algorithm did not converge


If I use:
cds = estimateDispersions(cds, method="pooled-CR", modelFormula = ~individual + condition)

I also get the same warning message.

[dpryan]

Try upgrading to DESeq2, it has a number of improvements. Getting that warning for just one or two genes is unlikely to be a problem. Just have a look at the diagnostic dispersions plot to ensure that things generally went as planned.