Entering edit mode
2.9 years ago
mimA
▴
30
Hello,
I'm wondering if there's a way to calculate the optimal number of samples per condition for a RIP-seq experiment. I haven't come across any R-package or method that would do power calculations although there exists one for ChIP-seq but it's not a very straightforward approach:
Zuo C, Keleş S. A statistical framework for power calculations in ChIP-seq experiments. Bioinformatics. 2014
Does anyone know if there is a recommendation on sample size for RIP-seq?
Thank you!
The question is what you hope to get from this. Based on RNA-seq papers we know that for really decent power one would actually need tens of samples, maybe more depending on the experimental setup. This is (even for RNA-seq) often not feasable, for ChIP-seq it is clearly not, both in terms of time and money. I am not familiar with RIP-seq it is probably similar. As a rule of thumb, do as many as possible given the circumstances, at least three as a minimum if differential analysis is the goal. If your read the papers on power, e.g. from Schurch et al https://pubmed.ncbi.nlm.nih.gov/27022035/ you see that adding additional replicates especially at low n (so e.g. from 2 to 3, or from 3 to 4) brings notable gain in power, with the curves (see the figures in the paper) than flattening when n increases. That having said, three as a minimum, more if you can.