[BioC] Model simplification on microarray data

[Paolo Innocenti]

Dear list,

I am working with Affymetrix GeneChip gene expression arrays. I am 
fitting to every transcript a mixed model (using lme4) that looks like this:

lmer(Y ~ female.type*male.type+(1|female.pop)+(1|male.pop))

where Y is the transcript signal. When looking at the results, I 
realized that many of those terms are not interesting (namely the 
interaction and the second random effect). For most of the genes (but 
not all!), a simpler model works much better. The simplification would 
be carried out using likelihood ratio tests.

The question is, does it make sense to fit different models (more 
specifically, simpler version of the full model) for different genes? 
Would I still be able to compare the terms that are present for all the 
genes (for example when I have to adjust the p-value)?

Thanks,
paolo

-- 
Paolo Innocenti
Department of Animal Ecology, EBC
Uppsala University
Norbyvägen 18D
75236 Uppsala, Sweden