I have a RNAseq and a microarray dataset in which I am trying to determine whether there is an interaction between age and diagnosis. Using DESeq2 I did a likelihood ratio test. The full design was ~Dx + Age +Dx:Age + otherConfounders(sex,race…) and the reduced design was ~Dx + Age + otherConfounders(sex,race…). What is the correct way to replicate the LRT analysis on the microarray data using the limma package? I am unable to figure out the appropriate commands.

Best 

Sarven

In limma, you only need to create the full design matrix, not the reduced. The following code will give a table of the top DE genes for the interaction:

design <- model.matrix(~Dx + Age + Dx:Age + otherCounders)
fit <- lmFit(y, design)
fit <- eBayes(fit)
topTable(fit, coef=4)

The reason for "coef=4" is that the interaction term is the 4th column of the design matrix. (The intercept is first, then Dx, then Age, then Dx:Age.)

Also useful is

decideTests(fit)

which gives the number of DE genes for every column of the design matrix.