I am a post doc student and I sent a lot tibias rats samples to a genome center this year for Microarray (Affymetrix ST 2.0). We did not polled any sample. We wrote a paper and now the reviewers are asking why only n=3. My questions are; a) How efficient and sensitive the affymetrix microarray are? b) Since we did not polled, do you think n=3 is sufficient for a reliable analysis?
it is true that n=3 is low regardless of pooled or not. This may affect the power you have to find differential expressed genes i.e. the reliability in terms of DGEs you have missed. I cannot comment on whether your experiment have enough power as it depends of your experiment and the effect of the treatment. It is possible your experiment has decent power if the effect is large and the experiment is tightly controlled.
It does not, however, put into question the reliability of the your discoveries (the DGEs you have found) as long have you controlled the type 1 error properly (with pvalues and multiple test correction methods). On top of that, the Affy arrays are good assay and one of the most sensitive array currently available.
Thus, I do not see the low number of replicates a critical design flaw. A lot of experiments actually are in the 2-5 range.