[BioC] Interaction terms on a factorial design without a reference sample

Bernardo Lemos blemos at oeb.harvard.edu
Wed May 23 04:56:02 CEST 2007

Hi all,

I am analyzing an experiment with two-color arrays in which  
expression data was collected from two genotypes in two different  
environments according to a loop design. I wanted to use LIMMA to  
test for effects due to genotype, environment, and their interaction.  
The LIMMA user's guide has a good example on a similar factorial  
design which includes, in addition to the four treatments of  
interest, an external fifth RNA source used as a reference. I don't  
have such a reference in my design and am having a hard time coding  
for the design matrix such that I can make the contrasts between  
genotypes (1&2), between environments (1&2), and also have an term  
for the interaction of genotypes and environments.

Below is a scheme of the design, with lines being dye-swap  

genotype_1_Environment_1 ---------- genotype_1_Environment_2
			|								|
       			|								|
			|								|
			|								|
genotype_2_Environment_1 ---------- genotype_2_Environment_2

Any ideas on how to code for the design matrix in such a case?


Bernardo Lemos.

More information about the Bioconductor mailing list