**40**wrote:

Hello, I apologize in advance for the dumb question but I am no expert in statistics, my course only covered how to do student's t.test pretty much. I could not find information in different forums on a similar problem specifically. I am not sure what to do in the following situation. Suppose we measure the expression of geneX by RT-qPCR in two different cell lines, each of which is treated with some drug or not treated. They are independent, so it is not the same cells before an after treatment; it is different cells that are either treated or not. Suppose we have 3 biological replicates for each condition, and the following values

Cell line A, no treatment: c(0.9,1,1.1) => mean expression 1 +/- 0.1 standard deviation (let's call this A1) Cell line A, treatment: c(9,10,11) => mean expression 10 +/- 1 sd (A2) Cell line B: no treatment: c(2.9,3,3.1) => mean expression 3 +-/ 0.1 sd (B1) Cell line B: treatment: c(290,300,310) => mean expression 300 +-/ 10 sd (B2)

Obviously, the drug causes an increase in both cell lines, but in line A there is a 10-fold increase (A2/A1=10/1) while in line B there is a 100 fold increase (B2/B1=300/3). I want to claim that the effect in B is much stronger.

What statistical test should we use to tell whether the RATIOS (gene expression upon treatment normalized to basal gene expression without treatment) are significantly different i.e. A2/A1 +/- propagated error != B2/B1 +/- error

Thanks in advance