I searched the web and found no resources for t-test or other similar tests on fold changes calculated using Pfaffl method. How to compare two groups or even one treatment and one control group?
I am sorry to have to bother you again but I saw that you posted here too :)
I suppose most of my confusion is stemming from the fact that I used the Pfaffl method for relative quantification which in the end provides a "transcript ratio" or "fold change" (target/ref). So for performing statistical tests (whether t test or anova or whatever depending on experimental design) you say here to use the fold changes but to take the log? but what base log would you use for fold changes derived using Pfaffl?
yes, you should use the log fold-change to do the stats. The base does not matter, as this will just be an offset to all the values, what cancelles out when you analyze differences. I would intuitively prefer log to base 2, as this is common in the transcriptomics field.
Jochen Wilhelm so finally in the end do the stats on the log2 of whatever the result of the following pfaffl equation is :
Transcript ratio: Etarget gene ^ (Ct for target gene in control sample - Ct for target gene in test sample) / Ereference gene ^ (Ct reference gene in control sample - Ct reference gene in test sample)
My setup consistis of 3 indepentent repeats/experiments with each 1 control group (reference) and 2 treatment groups. N is not always equal per group and also differs per experiment (N= 5, 7 or 8 per group).
When I perform the pfaffl method per experiment/repeat (I cannot already combine the 3 independent experiments at this point, right?), the averages of the delta Ct values both for the HKG and GOI for the reference group equal 0 exactly, the averages of the corresponding E^delta Ct s do not equal 1 exactly. Hence, the average of the expression ratios for the reference group range from 1 tot 1.3 (not 1 exactly).
Is it correct to perform statistics on the log2 transformed ratios (so taking into account all datapoints not just the averages) that I obtained per experiment? I am struggling because I was thinking of making a log2 fold expression ratio graph with the averages per experiment indicated but then the average log2 fold expression for the reference group will not be 0, which I thought is the point of performing this pfaffl method?
For the statistics I would put all the ratios (for each group and each experiment) in a linear mixed model correcting for multiple testing.