Using the qPCR procedure, this is well established how to calculate a fold-change for an individual sample: 2^-ddCt = 2^-dCt(tissue1) / 2^-dCt(tissue2) which is equal to = 2^(-dCt(tissue1) - (-dCt(tissue2))). But, should the mean fold-change be calculated as (1) a mean for all individual fold-changes of all the subjects or rather (2) a ratio of mean 2^-dCt(target gene) and mean 2^-dCt(reference gene)? These 2 methods provide different results due to the nature of mean value. What to do if tissue1 is not paired with tissue2 (e.g. they are not from the same patient - e.g. serum miRNA from different groups of patients)? Should I rely on method (2) then? What to do with confidence intervals (CI) for the mean fold-change then? Should I use Fieller's theorem?
I found that the CI for the ratio (method 2) can be easily calculated on-line with the method by Fieller:
https://www.graphpad.com/support/faq/how-can-i-compute-the-95-confidence-interval-of-the-ratio-of-two-values-knowing-their-sd-and-assuming-they-follow-a-gaussian-distribution/
Thank you for your answer!
Can I just generalize that the mean should be calculated with the data of normal distribution: -dCt, 2^-dCt, or 2^-ddCt? And then - if necessary - use the means in order to calculate the the fold-change (in case of means of -dCt and 2-dCt only)? This is reasonable as the arithmetic mean is really a good estimator only in case of symmetric probability distribution (e.g. normal).
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biomolecules
- Nucleic Acids
- RNA
- Small RNA
- microRNA