--Hi,
such as RTqPCR to analyse gene expression, is it necessary to have biological replicates in a ddPCR experiment ?
thank yu --
Did you mean technical replicates instead of biological replicates?
Hello Laurent Manchon
Yes, it will be required in a ddPCR experiment as well because biological replicates are biologically distinct samples (for instance, the same type of organism treated or grown in the same conditions), which show biological variation.
In ddPCR analysis, due to large number of droplets in each tube, each of which is an independent assay, there will be no need for technical replicates, as you know that technical replicates are repeated measurements of a sample, which show variation of the measuring equipment and protocol.
Best.
Well I agree & disagree with parts of the above comments. If the copy numbers are abundant/high, then technical replicates are redundant. But the current generation of dPCR/ddPCR instruments doesn't necessarily permit robust results for MRD >5.5/6.0-ish levels of sensitivity (i.e. for BCR-ABL). Hence sometimes you need to use 2-3 wells to ensure high sensitivity for some MRD-related applications (https://europepmc.org/article/pmc/pmc7710259). However, as the partitions/chambers increase we will see that we can use a single replicate to achieve what was once achieved with 2-3 replicates (i.e. for MRD). So the answer is contextual as well.
In that case, you would need also need biological replicates for statistical comparison purposes. Even a student's T-test need a few biological replicate to make a robust comparison. This is a nuanced topic that involves power analysis with some estimated parameters that you think you may observe with your study (https://statisticsbyjim.com/hypothesis-testing/sample-size-power-analysis/). It involves thinking about type I & II errors in relation to effect size(s). In general, bigger effect sizes need less samples (if produced with low(er) noise), whereas smaller effect sizes need more samples to show that they are not noise coming from the sampling/assay. For example, for a control vs. treated assay how confident would you be of a result that showed 5 Ct difference vs one that showed 0.05 Ct difference? You don't need as many samples to show that there is a (statistically) meaningful difference with 5 Ct difference, but with a 0.05 Ct difference you would need a lot of samples to make sure that isn't a type I error, or even a type II error by not being able to detect it. To conduct a stastical analysis, you absolutely need biological replicates to make inferences. Power studies help guide the number of samples with some rational, to not overdo it (when 10 is OK instead of 100) or underdo it (for Type II errors - lack of sensitivity). Hope that helps.
it's strange it's not mentioned in dPCR's MIQE (https://academic.oup.com/clinchem/article/66/8/1012/5880117)
- Badges
- Science topic
- Similar topics
- Biological Science
- Molecular Biology
- Molecular Methods
- Cloning