Hi there,
I'm working on a metatranscriptomics project and want to perform DEA using DESeq2. I am however not sure how to insert my design.
The sample set-up is as follows:
There are 3 plots on dry, 3 on wet ground. From each plot samples are taken from the top layer and deeper. Each plot has been sampled in spring, summer and winter.
My intuition on GLM's tells me to make a design formula that is either of these two:
design = ~ layer + moisture + season
or
design = ~ season + moisture + layer + season:moisture + season:layer + moisture:layer + layer:moisture:season
The first one delivers roughly 41 and 17% significant (KEGG) genes respectively up- and downregulated (when looking at top-deep, the strongest effect), while the second one only yields a fraction of that: 4,5 and 4,5%. Which method is the way to go to portray reality the best? And how do I interpret the big differences between the two methods? Why does it differ so much?
Thanks in advance!
Seppe
