[BioC] limma design: paired samples, two groups

[Jussi Paananen]

Hi!

I have somewhat basic question about limma analysis design. The 
experimental design is as follows (using Affymetrix single channel arrays):

1. We have a set of people, a baseline sample is taken from everyone 
(Control)
2. People are divided in two intervention groups (group A and B)
3. After intervention, sample is taken from everyone (After)

So, we basically have paired samples (before and after intervention 
Control/After) and two distinct groups (different types of intervention, 
A/B).

I am interested in the following questions:

1. Differentially expressed genes in group A between Control and After
2. Differentially expressed genes in group B between Control and After
3. Differentially expressed genes between Control and After

The first two cases are pretty simple; I can for example divide the 
expression data set to groups A and B, and then perform basic paired 
sample analysis inside the group between Control and After.

What I am not so sure about is that with the 3rd question, how 
could/should I take the group in to account with the analysis?

Any help would be appreciated! To provide a practical example, here is 
an example target frame:

   FileName Person    Time Group
1    1.cel      1 Control     A
2    2.cel      1   After     A
3    3.cel      2 Control     A
4    4.cel      2   After     A
5    5.cel      3 Control     B
6    6.cel      3   After     B
7    7.cel      4 Control     B
8    8.cel      4   After     B

And this would create a design matrix for a basic paired analysis 
between Control/After:

 > person <- factor(targets$Person)
 > group <- factor(targets$Group, levels=c("A","B"))
 > time <- factor(targets$Time, levels=c("Control","After"))
 >
 > design <- model.matrix(~person+time)
 > design
   (Intercept) person2 person3 person4 timeAfter
1           1       0       0       0         0
2           1       0       0       0         1
3           1       1       0       0         0
4           1       1       0       0         1
5           1       0       1       0         0
6           1       0       1       0         1
7           1       0       0       1         0
8           1       0       0       1         1

Thanks,
Jussi

-- 
Jussi Paananen