Hello

Seen similar questions but got no response on older threads so just looking for some guidance.

Trying to run an lrt using the deseq tutorials. Have done it before with a different experiment, but wanted to check that I have understood the full and reduced model in order to examine the effect of the correct variable.

The experiment involved two conditions and a control, the effect of the conditions (and control) were evaluated over time (day 0, 1, 3 & 7). I would like to look gene changes over time for each condition (I/e which genes and with what clusters are seen for each condition and control).

Thus the experiment looks like this (the RNA is from mouse cells)

Condition 1 : Sample Day 0, Sample Day 1, Sample Day 3, Sample Day 7

Condition 2 : Sample Day 0, Sample Day 1, Sample Day 3, Sample Day 7

Control : Sample Day 0, Sample Day 1, Sample Day 3, Sample Day 7

If 'intensity' relates to the conditions/controls. and Day refers to the time point. Would the correct full model be:

design = ~ Day + Intensity + Day:Intensity ---> this is full model

and the reduced model to evaluate the effect of 'time' on the expression for each intensity condition be:

dds_lrt <- DESeq(dds_filtered, test = "LRT", reduced = ~ Day)

Thanks in advance

BW Rob

When you perform an LRT, you are testing the coefficients that are no longer included in the reduced model. So you are testing to see if including Intensity and the Day:Intensity interaction fits the data better than a simple model that only includes Day.