Hi All! In the bioreactor field the DOE method is used, where factors are mixed creating blocks. The goal of this analysis is to find experimental designs for investigating the dependence of some measured quantity (gene expression) on a number of independent variables (factors), each taking different values (levels), in such a way as to minimize the variance of the estimates of these dependencies using a limited number of experiments. This is also called factorial design. Something like this table:
Now as you can notice here factor A is measured 4 times for each value (condition 1 and condition 2), but always with the contemporary effect of another factor.
The DOE allows to do a fitting of a model and estimate the effect of each factor on the observed variable.
Now, my question is: if I wanted to add an RNA-seq experiment on all the 8 samples, how many biological replicates shall I have for each of them to get enough statistical power to say something about a gene and the factors A,B,C and D (and maybe also of the combinations as in the table?).
I know of R packages such as edgeR able to deal with these design, but I am not sure how I can really benefit from this design and what is the role of biological replicates in increasing this power.
Thank you!
