In gene expression experiments if I have a replicate data where target and reference are amplified in separate wells, how can I calculate SE for ddCt if I have mean dCt for control (one value) and mean dCt for experimental (one value)? and should I use Ct, ∆Ct or ∆∆Ct to calculate the interval plot ??
By error propagation:
https://en.wikipedia.org/wiki/Propagation_of_uncertainty
For the difference of two mean ∆Ct values with SEA and SEB the SE(B-A) is:
sqrt( SEA²+SEB² ).
However, it is recommended to get the SE of the ∆∆Ct from the linear model on the ∆Ct values, because the error propagation tends to overestimate the resulting uncertainty. But if you only know the mean ∆Ct values per group but not the individual ∆Ct values, there is no other chance. If you don't know the SEs of the mean ∆Ct values either, there is no way to get the SE of the ∆∆Ct.
Dr. Wilhelm
Excuse my ignorance, in the table (attached file), why did they recalculate the SD while calculating ∆∆Ct ??
Thank you for your help so far
I have no idea what they did there. The SE of the ddCt for these data is 0.082.
Apart from that I have 2 problems:
- the data looks too homogeneous. I have never seen such small variances between independent experiments. I suspect that these were not really independent experiements.
- using (only) GAPDH as a reference between different cell types is risky. I would not assume that the GAPDH expression is identical between liver and kidney. If the experiment is well standardized (as it looks by the extremely low variance) one would expect similar ct values for GAPDH in both organs, but there seems to be a systematic difference.
That's all strange. I wouldnt trust any conclusion unless I knew how the experiment was performed and how the reference gene was validated.
Dr. Wilhelm
I have a small number of single cell oocyte for a controls and samples and I couldn't collect more of them for obvious reasons, now in gene expression experiment using qPCR the values for Ct mean is quite promising yet the SD for Ct (because of (#n)) is a large number =>SD (dCt) is bigger => SD (ddCt) is more and more, that will affect the rang of data for folding (2^(-ddCt) for sure, do you have any suggestion for such a dilemma?
Thank you.
If you can't increase the sample size, then there are only two options: use a different method or change your question.
- Badges
- Science method