I want to make a GE analysis. I have the expression datasets (Human Gene 1.0 ST) CaseA1, CaseA2, CaseA3 which are very similar and the datasets CaseB1, CaseB2, CaseB3. My aim is to identify up or downregulated genes (log ratios). What is the best way to do this? Compare e.g. CaseA1 with CaseB1 then CaseA2 with CaseB2 and then CaseA3 and CaseB3. Or build the mean of CaseA and compare it with the mean of CaseB (plus multiple testing correction)? Thank you!
The normal approach to a two-group comparison with replicates is to use a statistical test, often a variant of a t-test, applied to each gene to look for differences between the groups. Whether that test needs to be a paired test is dependent on how the samples are related to each other.
Any software package that purports to be capable of gene expression analysis will be able to perform this computation. There are many commercial packages as well as open source solutions. To get started, I would suggest finding someone local who has done microarray gene expression analysis to work with you on your data until you are comfortable with working on your own.