Clarification on BY2005.R script

[jschreie]

Hi, I have been looking for functions to perform multiple corrections for confidence intervals and found your SelectiveCI scripts. I have bootstrap estimates of mean gene measurements, have constructed 95% confidence intervals, and have compared intervals from an experimental treatment to a control treatment, considering those genes with non-overlapping intervals to be statistically significant.

If I understand the BY2005.R script correctly, I would input x=mean estimate, sigsq=sd of the population squared?, n.rejections= number of of genes with overlapping intervals?, n.tests=number of genes tested?, alpha=0.05

Is the above an appropriate use of this function for generating FCR adjusted multiple confidence intervals?

Thanks,
Jeremy

[johnros]

Have a look at https://github.com/johnros/selectiveCI/blob/master/selectiveCI/R/BY2005.R
You may see that sigsq is the standard error of the mean in the CI formula.
(I guess I did choose a confusing name for that variable. I don't recall why that is).