[BioC] Opinion on proceeding strategy

[Marco Gentilini]

Hi everybody
             I'm using Affymetrix Genechip and I was wandering if it is
reasonable to use a dataset whose genes belong to different analysis of the
same experiment (a comparative analysis with three replicates). The genes
belonging to all the analysis would compose the dataset to use for further
analysis. I'd use the same bg correction and pm correct methods (RMA and
subtractmm rispectively) but different normalization and summarization
methods. 
For example 1st analysis bg correction :RMA
                         normalization :loess
                         pm correct    :subtractum
                         summarization :medianpolish
            2nd analysis bg correction :RMA
                         normalization :invariantset
                         pm correct    :subtractum
                         summarization :liwong
            3rd analysis bg correction :RMA ...

If anybody could tell me what do you think about it I'd be very greatful.
Thanks everybody