View Single Post
Old 09-16-2015, 08:25 AM   #5
Devon Ryan
Location: Freiburg, Germany

Join Date: Jul 2011
Posts: 3,480

Ah, I had missed that you explicitly requested an LRT with DESeq(). The default is Wald there too. The models that you're testing are completely different between the two methods. With a Wald test, you can test individual coefficient, with the LRT you're testing a more generic interaction.

Edit: Yes, an LRT is likely what you want, since it asks the question, "what genes generally interact in a time:condition dependent manner, regardless of the exact time point".

Last edited by dpryan; 09-16-2015 at 08:39 AM.
dpryan is offline   Reply With Quote