I would like to understand calculations made by qPCR software. I create a standard curve in Excel with same parameters as in software with equation for my target: y=-1.571ln(x)+39.433 and calculate CN from exactly same Ct (same decimal places).
x (CN)=((y(Ct)-39,433)/1.571)^e; where (e=2.71828183), and after normalization against used amount (ng) and genome size, I get average CN=8.69, but software result is 8.71.
Then I try equation used in the software : Ct=m[log(Qty)]+b, from that x=10^((Ct-b)/m), where b is y-intercept (= 39.433 in this case); m=slope. I get similar results, but not exactly the same average: CN=8.72.
where I'm wrong? are there additional corrections or coefficients usually used in the qPCR calculations by software for absolute quantification?
Hi again Angelika,
variation of 100 copies out of a total of 4000 (or more) represents 2.5% (or less)... Is this not acceptable? Don't forget quantification is made from experimental Ct data input on a linear y-scale transferred onto a logarithmic x-axis (so any experimental error on Ct becomes exponential on CN: this is the 2.5% variability on your CN). To illustrate, let's say you calculated once CN=4000 copies, it means you actually measured Ct= -1.571*(ln4000)+39.433=26.403. If CN was 4100, it means Ct was 26.364... How different are 26.403 and 26.364? It's around 0.1%...
Finally, If your data comes from different plates, first you have to normalize the results from each plate (towards internal standard) before any further calculation.
Hi there,
I don't see what your problem is: 8.69 is very close from both 8.71 and 8.72! You should be very happy with it.
Hi, Dominique,
I was satisfied until I calculate results for the target with more than 4 000 copies per genome, difference grows to 101.35 copies. Ok, averages are still close, but I would like to know exact formula for calculations as I need to combine several plates of results... There are enough variables in the experiment, I don't like to introduce additional variation due to the math.
I just would like to know opinion of experienced people. Are my calculations correct or I lose any valuable parameter in the middle; or I can introduce this type of error by choosing alternative formula in excel.
Hi again Angelika,
variation of 100 copies out of a total of 4000 (or more) represents 2.5% (or less)... Is this not acceptable? Don't forget quantification is made from experimental Ct data input on a linear y-scale transferred onto a logarithmic x-axis (so any experimental error on Ct becomes exponential on CN: this is the 2.5% variability on your CN). To illustrate, let's say you calculated once CN=4000 copies, it means you actually measured Ct= -1.571*(ln4000)+39.433=26.403. If CN was 4100, it means Ct was 26.364... How different are 26.403 and 26.364? It's around 0.1%...
Finally, If your data comes from different plates, first you have to normalize the results from each plate (towards internal standard) before any further calculation.
Thank you, Dominique.
That's what we are thinking about these variation. I will be somewhat more convinced of the results now. We will normalize the result from different plates as I have placed identical sample in triplicate for this purpose in each plate. Also measurement threshold for each target at different plates is identical.
- Badges
- Science method