Hello, I'm having trouble constructing a proper design matrix for statistical analyses.

I am working with 2-color Agilent array data for multiple arrays. I've bg-corrected and normalized the expression values.

However, the design of the arrays are not uniform.

The goal of the experiment is to see if there's differential expression between pre-treatment and post-treatment with a drug.

However, the microarrays were formatted as follows:

pre = pre- drug treatment
post = post- drug treatment 

                 file           Cy3               Cy5
Array 1      file1.txt          pre-sample1       pre-sample2
Array 2      file2.txt          pre-sample3       post-sample3
Array 3      file3.txt          pre-sample4       pre-sample5
Array 4      file4.txt          post-sample6      pre-sample6
Array 5      file5.txt          post-sample7      pre-sample7
Array 6      file6.txt          pre-sample8       post-sample8

I've been following this guide http://koti.mbnet.fi/tuimala/oppaat/r2.pdf and I'm a bit lost to what the design matrix would look entail.

Is there any way to design a matrix with the pre to pre arrays in there or do I need to repeat the analysis twice, one analysis with just Array 1 & 3, a second analysis with the remaining groups?

Any guidance would be appreciated.

edit: is design <- modelMatrix(targets, ref="pre") an accurate design matrix?