hi, i was wondering which is the best way to grouping GSM from independent experiments and analyze them considering interexperimental variability also related to different platforms used. After gene mapping some have NAs and some have very high expression that is not in line with the rest of the group.

I start thinking it is not a reasonable analisys though.

The only defensible way is to make comparisons within experiments and then do a meta-analysis. But that assumes you have identified multiple experiments on GEO that each have the same type of samples for the first step.