Removing batch effect from GEO database

Question

[eric.palanques]

Hi,

I have downloaded several GSE from GEO database, all from Homo sapiens and from the same platform and tissue. I have merged all the datasets and performed RMA normalization on the merged set. However, I can still see a lot of unexpected variation between them. I suppose it is due to batch effect. How could I remove it?

Thanks in advance,

Answer 1

[Kevin Blighe]

If you believe there are 'extraneous' effects / unwanted sources of variation in the dataset that could bias your results, then you could perform Surrogate Variable Analysis, which will essentially identify a set of variabes that 'capture' this unwanted variation. The idea, then, is to include these variables in your design forumla for the purpose of controlling for these.

Kevin

[eric.palanques]

Thank you very much, that seems to work