Analysis of ratios (fold changes) from different microarray experiments with common control

Question

[matt.arno]

Hi - I have two gene expression experiments that I performed on microarrays at different times:

expt 1: 3x untreated controls 3x treatment A

expt 2: 3x untreated controls 3x treatment B

note: the untreated controls are from the same RNA in both experiments.

Whilst I can easily determine treatment A and B effects separately, I would also like to find out what is statistically different between A and B. I can do a Venn diagram to see what is unique and common to both lists, but I was wondering whether there was a more statistical method.

I thought that perhaps I could normalise the treatment signals to the relative average control signal in each case, and then compare these ratios (A/CTRL vs B/CTRL) to each other in a t-test or similar.

Or make a joint experiment with all 12 samples and somehow construct a linear model to take into account the control groups measured twice.

Thanks for any help

Cheers, matt

[WouterDeCoster]

I guess you could get an idea of (severity of) potential batch effects by doing the comparison of the control samples in both runs against each other.