I am analysing some microarray data. For a set of participants, I have their RNA expression before an intervention ("Baseline") and after the intervention ("Followup"). I want to know which genes are differentially expressed between the two timepoints. For this, I am using a moderated paired t-test, as outlined in the Limma user guide:

Time <- factor(y$targets$Time,levels = c("Baseline","Followup"))
Participant <- factor(y$targets$Participant)
design <- model.matrix(~Participant+Time)
fit <- lmFit(y,design)
fit <- eBayes(fit)
topTable(fit,coef="TimeFollowup")

Next, I would like to see if adjusting for other factors affects the differential expressions of genes. I have other information about the participants in y$targets, including age, sex, etc. How can I modify my model.matrix to adjust for these factors? E.g., if there was a gene which was differentially expressed in men but not women, how would I know? Or, if there was a linear relationship between age and differential expression of a gene?

Let me know if any more info is needed, really appreciate any help!

Hi Emma,

The standard procedure would be to include other covariates in the design formula. So, for example, one could use:

design <- model.matrix( ~ Participant + Time + age)

The effects of Participant and age will then be taken into account when you derive test statistics for Time. If, however, you are already adjusting for the Participant effects, then, perhaps, introducing more covariates could prove problematic for the model fitting.

There is further information that elaborates on this —and also on the use of blocking factors via duplicateCorrelation()— here: limma blocking and including covariates

Kevin