EdgeR repeated measurment

[teshome]

Hi everybody,
I have question on repeated measure analysis using edegR or DESeq.
The experiment contains two groups (low and good) and three time point(0 day, 14 day, and 28 day) from 6 individuals in each group (i.e. repeated in each time; in total 36 samples). I want to check which gene is differentially expressed within the group between time points and between groups at different time point.

I am think to analyse using a mixed-model nested repeated measures.
where;
quality = main effect (i.e. good or low - fixed)
individuals(quality) = individual nested within quality (random factor)
time = fixed factor and repeated factor
time x quality = fixed factor
time x individual(quality) = random factor

Is this sound ok? If so, can I model it in edgeR as it is mentioned in section 3 of user guide, like this:

quality <- factor(DATA$quality, levels=c("good", "low"))
time <- factor(DATA$time, levels=c("0day","14day","28day"))

design <- model.matrix(~quality+quality:individual+quality:time)

How the contrast argument works for example to compare good:0day vs low:0day? Should I relevel the base to 14 day or 28 day to get the contrast among time points?
Thanks

[dpryan]

Neither edgeR nor DESeq(2) support mixed-effect models; the normal way to handle this would be to simply set individual as a factor. BTW, "quality:individual" doesn't make any sense since each individual is in only 1 group. Just use individual alone. The model matrix you're looking for is simply:

design <- model.matrix(~individual + quality*time)

I'm assuming that there may be a time-effect, but if not just remove that main effect.

[teshome]

Thanks Ryan,
I forgot to mention I blocked individuals, that is the reason "quality:individual" in the model. I am not interested in individuals per se rather their variance component.