I've performed experiments to analyze the effect of different concentrations of a drug on gene expression using RT-qPCR. It is not clear to me whether data should be ploted as Fold Change vs. log(conc) or as log(FC) vs. log(conc). Also, if I want to show the expression as the percentage of the maximal response, which values -FCs or log(FC)- are to be used and which is the correct way to calculate it.
Any further advice is welcome.
Thank you
This strikes me as a problem better answered by actually considering your data, rather than establishing in advance what method you should use. There is no "one size fits all" answer: the expression response of a transcription factor that goes from "basically none" to "loads" would use a different plot from, say, a myosin heavy chain that increases from "loads" to "1.2x loads".
For the former, you'd use a log plot, because the data would look silly any other way. For the latter, you probably wouldn't, because while a 1.2-fold increase in something as massively abundant as a myosin heavy chain absolutely IS of biological relevance, it would appear to be 'no change at all' if plotted on a log scale.
If your response is so dramatic (i.e. orders of magnitude difference) that it is best conveyed by a log plot, convey it with a log plot. If it isn't, then...don't.
As for percentages, use the actual values, not the log transformed values. The percentage of a log transformed value is...meaningless, really.
Matías Gabrielli
If your statistical analyses were on dCT (or ddCT) values your error bar in FC plot would be asymmetric. Therefor, you should log scale in y axis to achieve symmetric error bar of FC.
ddCT is the differnce between refrence gene and target gene, so you will calculate the fold change between them
Good Luck !
Not quite. What you describe are dCt values (the difference between two Ct values). The ddCt is the difference between two dCt values. The dCt value indicates -up to an unknown, assay-specific, additive constant- the "log expression" of a gene, normalized to the expression of a reference gene. The ddCt is the difference of two such "log expressions", typically between two different conditions, and indicates tha log fold-change or log fold-difference of the gene expression between these two conditions (the "unknown constant" that is included in the dCt values cancels out when dCt values are subtracted from the same assays).
Its a good question and unfortunately none of the previous answers really deals with what you are asking in my opinion. While I think we are all on the same page that one should calculate your error and do statistics on the Log2 data, is it correct to keep your data on the log2 scale when doing a dose-response curve fit?
In case you want do calculate an IC50 from your dose response, you will get a different IC50 depending on whether you fit your curve to the dCt and ddCt datapoints (on the Log2 scale) or when you first convert your Log2 values to a linear scale (e.g. fold-change or %response). Typically your IC50 will be lower when using the linearized values (2^dCt or 2^ddCt).
I would also be very interested in hearing what the "right" way is to show dose-response curves with IC50 determinations from qPCR data. Log2 data or linear data ?
In case you want to calculate your %inhibition or response i use the following formula:
%inh = 100x(Sx - So)/(Sc - So)
Sx = 2^ddCt of your sample with a given concentration of your compound
So= 2^ddCt of your negative control sample without compound (stimulated cells without compound, 0% inhibition)
Sc= 2^ddCt of your positive control sample (mock treated cells, 100%inhibition)
%Inh: Capacity of your compound to inhibit whatever stimulus you gave your cells
- Badges
- Science topic
- Similar topics
- Biological Science
- Genetics
- Gene Expression