[BioC] LIMMA: paired samples and common reference?

[Enx Enx]

Hello,
 
I havea an experiment with mice done on Affymetrix and it looks like this:
 
FileName   Treatment   Batch   MouseID
===============================
file1            control        1           101
file2            control        1           102
file3            control        1           103
file4            treatA         1           104
file5            treatA         1           105
file6            treatA         1           106

file7            control        2           107
file8            control        2           108
file9            control        2           109
file10           treatA        2           110
file11           treatA        2           111
file12           treatA        2           112

file13            control        3           113
file14            control        3           114
file15            control        3           115
file16            treatB         3           116
file17            treatB         3           117
file18            treatB         3           118

file19            control        4           119
file20            control        4           120
file21            control        4           121
file22            treatB         4           122
file23            treatB         4           123
file24            treatB         4           124
 
The controls from all batches are (should be) biologically the same and I am looking for differentially expressed genes in:
a) treatB versus treatA, 
b) treatA versus control,
c) treatB versus control.
 
Are the design matrix and contrast matrix as follows?
 
factor_treatment = factor(Treatment, levels=C("control","treatA","treatB")) 
factor_batch = factor(Batch, levels=C("1","2","3","4")) 
design = model.matrix(~0 + factor_batch + factor_treatment) 
fit = lmFit(eset, design)
cont.matrix = makeContrast(
                  treatBvsA = (treatB - treatA),
                  treatAvsControl = (treatA - control),
                  treatBvsControl = (treatB - control),
                  levels = design)
fit2 = lmFit(fit, cont.matrix)
fit2 = eBayes(fit2)
 
Cheers,
Daniel Berger