Statistical test for relative quantification - (Jun/21/2006 )
Does anyone do statistical tests of their relative quantification data (i.e. ANOVA, t-tests)?
I am looking at two genes and eight treatments. I know if fold changes in one are significantly than another.
we typically do the t-test. I think tests for significance are a good idea if you want to publish the data.
however, we never publish only qPCR data...we always back it up with at least one other method
Aimikins - I am about to do my first qPCR and from it publish my first paper. At the moment I am going to do relative quantification with SYBR green.
Am interested in what you use to back up your qPCR data.
well, at least in our opinion, change in mRNA levels is not always significant, or relevant. you also need to show increase in protein, etc
for example, we recently published a paper with quite a bit of qPCR data. however, we also used IF to show increased protein; we used EMSA to show increased binding of transcription factors to our gene's promoter, and we did some western blots to show increased phosphorylation of a protein known to be involved in upregulation of our gene (we were unable to get a good western with our protein itself after several tries; sometimes you have to try another tactic...hence the IF)
sometimes the RNA level may go up and that doesn't necessarily mean you'll get more protein (think about all post-translational modification, etc). it all depends on what you are trying to prove in your paper. we wanted to show that our protein was induced by our treatment; we did not consider qPCR data enough to make this statement conclusively. the problem with qPCR, it is so sensitive to so many variables, it's hard to believe the data even when it's reproducible, if it isn't backed up using at least one more approach
but that is just our opinion. many people publish only qPCR data; again it depends on what you're trying to prove in your paper