[BioC] limma novice seeks guidance: 17 comparisons among 2 groups (size 4 & 8)

[Paul Shannon]

Though I have spent many hours reading the limma user's guide, and pored
over chapter 23 of the Gentleman, Carey et al Bioconductor book, I 
have many more questions than answers regarding limma, and its use with
our data.   I come from a programming
background, with some math training (in the distant past); I am sure
many crucial statistical issues elude me.   I am looking for help.

Our experiment seeks to discover Plasmodium falciparum up-regulated genes
in peripheral blood samples drawn from four preganant women and eight children,
all of whom have malaria.  A schematic of the design may be seen at 

    http://gaggle.systemsbiology.net/pshannon/limma.html

In this diagram, each line represents two slides which are simple dye-swap pairs.
Due to a shortage of mRNA, only some of the many possible comparisons were made.

I have performed some separate limma analyses of each dye-swapped pair.
Many of the same genes -- the top 10 or so -- appear in most slides. These
genes make sense biologically.

Now I wish to probe more deeply, analyzing the data in aggregate to
discover more from the data.

In the BioC book, sections 23.2 ('Data representations') and section 23.9
('Direct two-color designs') suggest to me that we have a direct 2-color
design.  As a first step, I hope to find out if that is true.  If so --
or if not -- I am sure the answer will lead to more questions...

Thank you,

 - Paul Shannon
   Seattle Biomedical Research Institute & Institute for Systems Biology