[BiO BB] GeneExpressionOmnibus(GEO) question

[Ilya Venger]

Hi,
I am in the process of building a local expression database for  S. 
cereveisiae. The best source of expression data I could find was GEO. 
They state there that the data they present in GDS accessions is 
normalized, but they don't explain how.
The GSMs of different contributors not necessarily match, and they 
sometimes lack parts of the data. Also, if the raw data is submitted, 
how come they don't show the measurements for both repetitions of each 
gene in a cDNA array.
Another problem, is that although the GDS are claimed to be normalized 
the sum of logs for all genes on a chip doesn't sum to 0.

So, if anybody worked with GEO before, and has experience at it I would 
be glad to pose several more specific questions.

Cheers,
Ilya.