Hua Xiao, great comment, absolutely right. The level of RNA is only a very crude and possibly quite wrong measure of transcriptional activity. The latter is hard to measure directly. One would need to measure nascent RNA molecules (e.g. unspliced variants, or by incorporation of spiked radioactive nucleotides). At best, it gives a kind of an average of the gene activity, given there is a steady-state of transcription and degradation. And going further, one should not automatically set the amount or regulation of mRNAs equal with the amount or regulation of proteins. It may be similar, but often it is entirely different. The transcriptional network does not 1:1 map to the protein expression network. These are two different layers of kybernetic systems; they are linked, but one is not a blueprint of the other.

"do i use a ratio of the absolute number of gene of interest over the reference?" - yes, exactly.

" my issue is that biologically speaking, cells don't really respond to absolute number change but rather a fold change, which is more accurately presented in the logarithmic value given by the qPCR not by the actual value. " - yes!

Almost all biology "works on the log scale". It's almost always the relative changes (not the absolute) that are of relevance, and relative changes can be understood and analyzed well with standard tools when using their logarithms. It's a pity that so few biologists apprechiate the logarithm (many are rather scared of it). And it seem to be a major problem that Ct values from real-time PCR are already measures on a logarithmic scale of expression.

However, you are completely on the right track!

Because of problematic normalization with mRNA, a simple and easy way is to firstly create cDNA and then normalize your cDNA concentrations in your samples, measured with a NanoDrop device

When you determine and have the absolute number of transcripts/cell from a reference gene that is constantly expressed, you would immediately know the numbers of molecules from genes of your interest by comparing the relative levels. So, I absolutely agree with @Jochen, what is important and of relevance is the relative changes or relative levels of expression.

I would like to add to my above answer. I think there is a distinction between level of transcription and level of RNA. The latter can also be regulated post-transcriptionally through RNA stability.

Hua Xiao, great comment, absolutely right. The level of RNA is only a very crude and possibly quite wrong measure of transcriptional activity. The latter is hard to measure directly. One would need to measure nascent RNA molecules (e.g. unspliced variants, or by incorporation of spiked radioactive nucleotides). At best, it gives a kind of an average of the gene activity, given there is a steady-state of transcription and degradation. And going further, one should not automatically set the amount or regulation of mRNAs equal with the amount or regulation of proteins. It may be similar, but often it is entirely different. The transcriptional network does not 1:1 map to the protein expression network. These are two different layers of kybernetic systems; they are linked, but one is not a blueprint of the other.

Thank you Jochen Wilhelm . It helps to feel like I'm on the right track. but it's hard to make my PI understand this logarithm concept for some reason. my aim now is to somehow do this ratio of the absolute numbers I get from the Cq vlaues. However would you say that the fold change that comes from a ratio of absolute numbers is equal to that of logarithmic numbers? As far as my Math knowledge goes, that's a NO. But I would really like your input on this. Thanks again

Not quite sure what you mean.

The log of a ratio is equal to the difference of the logarithms:

log(A/B) = log(A) - log(B)

This you surely know... so what is your question?

i agree with https://www.researchgate.net/profile/Asadollah_Ahmadikhah2

Jochen Wilhelm I am exactly refering to this phenomenon actually. let me try as best I can to explain it. using the delta-delta Cq method we are able to evaluate fold change, but also we are able to evaluate fold change through the ratio of the absolute values resulting from Cq. how are these fold changes -if they are- comparable, as mathematically Log(A-B) is not equal to logA/logB?

Taking a ratio of (d)Cq values makes no sense, as these are already representing logarithms:

CtA = u - logE(NA)

where u is an assay-specific (unknown) constant, E is the (possibly unknown) amplification efficiency, and NA is the initial concentration of the target sequence "A" (in an arbitrary unit).*

The same formula for the reference gene "R", assuming the same amplification efficiency for A and R:

CtR = v - logE(NR)

Note that the unknown assay-specific constant is different, because these are different primers, possibly also a different probe.

The ratio of the initial concentrations NA/NR is obtained vis the difference of the Ct values. To see that more clearly, let's first solve the Ct formulas for the log(N)'s:

logE(NA) = u - CtA

logE(NR) = v - CtR

Now,

logE(NA) - logE(NR) = (u - CtA) - (v - CtR)

A difference in logs is the log of the ratio:

logE(NA / NR) = (u - v) - (CtA - CtR)

This step works only if we can assume equal amplification efficiencies of A and R. And here is where the dCt enters the game. I would naturally set

dCt = - (CtA - CtR) = CtR- CtA

so that

logE(NA / NR) = (u-v) + dCt

As you can see, the dCt is a log-ratio of concentrations, shifted by some unknown amount. If we knew E, we could anti-log (although, to my opinion, we shouldn't) to get

NA / NR = E[(u-v) + dCt] = E(u-v) * EdCt

Thus, the ratio of the initial concentrations is some unknown multiple of E to the power of dCt (that is, EdCt is proportional to that ratio, with an unknown proportionality factor). This ratio (hence: the dCt value) represents a measure for a "normalized expression" (the expression of A, normalized to that of R; the expression of R is used as a loading control). To simplify the following notation, I will use Q to indicate this ratio:

NA / NR = Q ---> logE(NA / NR) = log(Q)

This quantity can be compared to a similar quantity obtained in a different sample or under different experimental conditions. Let's say we like to compare this ratio in a "treated" sample to that in a "control" sample. We thus have a dCt from a treated and another from a control sample:

logE(Qtrt) = (u-v) + dCttrt

logE(Qctl) = (u-v) + dCtctl

The assay-specific but unknown constants u and v are the same, as the same assays were used for both samples. Again, the differences of the logs is the log of a ratio:

logE(Qtrt) - logE(Qctl) = [ (u-v) + dCttrt ] - [ (u-v) + dCtctl ]

Here, (u-v) cancels out and we get

logE(Qtrt) - logE(Qctl) = dCttrt - dCtctl = ddCt = logE(Qtrt/Qctl)

The ddCt is thus the log of a ratio of normalized expressions obtained under different conditions, ie.. a log fold-change of expression between these conditions. Again, if we knew E we could (but we shouldn't...!) antilog to get the fold-change

EddCt = Qtrt/Qctl

Note that I calculated the dCt different to the way it was defined in Livak's paper, therefor I don't get the minus in the exponent!

I can't make that clearer.

----

*This formula follows from the exponential amplification of a product in a PCR, where a fluorescence signal is generated that is proportional to the amount of amplicon in the PCR, and the Ct is the (fractional) cycle number at which this fluorescence exceeds some arbitrary threshold. The constant "u" includes all these assay-specific things like the proportionality factor between amount and signal, and the threshold value. Note the minus in front of the log: it says that the Ct value decreases with increasing concentration of the target sequence.

Jochen Wilhelm thank you very much for your time and insight. this is very helpful!

by using BestKeeper or REST software

So I would like to ask. How do you proceed? You do 2 standard ciurves, one with you gene of interest and one with your reference gene? How do you take the efficiency of the RT in your setup? In my hand I need to quantify a number of RNA copies of my gene of interest in tissue lysate. I don t really know how to normalize it. Thank you

@Isabelle Clerc What I'm doing currently is creating a standard curve with the absolute quantities measure by digital droplet PCR (ddPCR). then after quantifying my gene of interest and reference gene on their respective standard curves via RT-PCR, I do a simple fold change calculation. This method give's you a pretty good idea of the fold change in your RNA expression. the downsides are: it does not consider your PCR efficiency, It does not highly reflect the absolute number of RNA you have and will not be as accurately reproducible due to reverse transcription efficiency differences across experiments, and it's only a work around way that can be more effectively calculated by another method. To overcome these limitations, you can more effectively use the Pfaffl method of relative quantification, along with using a standard curve to find the efficiency of your qPCR, as that is required for the pfaffl method. of course your normalization utilizes the reference gene (your choice of reference gene has to be validated for stability with your experimental condition, but that's a different topic). I hope I answered some of your questions. if you need more clarifications, feel free to email me.

Very clear, thank you. But I don t understand the type of standard curves you use. Are these RNAs that have been reverse transcribed in the same time and conditions than your samples, or are you using directly DNA ? Thank you