qPCR Error Propagation: Subtraction of an arbitrary constant?

Question

[ebs15242]

I'm trying to understand how to present variation qPCR data when using the ddCT method.

I'm reading ABI's Guide to Performing Relative Quantitation of Gene Expression Using Real-Time Quantitative PCR that uses an example pulled directly from Livak 2001.

I follow how they propagate error through the Ct1-Ct2=dCt calculation, by squaring the standard deviations, summing them, and then taking the square root. That makes sense.

s = (s1^2 + s2^2)^(1/2)

But when they do the second subtration, they don't use the same formula. They say:

The calculation of ddCt involves subtraction of the dCt calibrator value. This is subtraction of an arbitrary constant, so the standard deviation of the ddCt value is the same as the standard deviation of the dCt value.

This makes no sense to me. How is the calibrator value an "arbitrary constant", when it's measured and has uncertainty? In their example, it has even more uncertainty than the experimental value.

Is this approach correct? Why do they not use the square root of the sum of the variances like they did the first time?

Answer 1

[Kevin Blighe]

The links that you posted just run into 454 errors? - these are the correct links:

• https://assets.thermofisher.com/TFS-Assets/LSG/manuals/cms_042380.pdf
• http://www.gene-quantification.net/livak-2001.pdf

I could only take a quick look. It seems that, in this manuscript, they are going over different strategies for normalising PCR data deriving from different experimental setups. The wording in the text that you quote —specifically the word "arbitrary"— is perhaps causing some confusion. When they say 'arbitrary', I think that they are referring to the mean of the Ct in each replicate group, which is essentially an 'arbnitrary' value that is assumed to represent the replicate group in question.

Just to link up another related answer: A: How to report and plot qPCR data

If still in doubt, I would contact the authors.

Kevin