find_peaks function significance 0.9 and 0.95

[Kareem Midlej]

Dear Karl,

I am writing to confirm the usage of the code I used to find the confidence intervals for significance levels of 90% and 95%.

The code is structured as follows: peaks_morphology_0.95 <- find_peaks(out, map, threshold=7, drop = summary(operm,alpha = 0.95))
peaks_morphology_0.90 <- find_peaks(out, map, threshold=7, drop = summary(operm,alpha = 0.90)) My question is whether the alpha value used in the summary(operm, alpha = 0.90) and summary(operm, alpha = 0.95) functions directly corresponds to the desired significance levels of 90% and 95%, respectively. Thank You! Best Regards, Kareem

[Karl Broman]

I would use something like summary(operm, alpha=0.05) to define the threshold to use in find_peaks, but I wouldn’t use it for the drop argument. The alpha argument in summary.scan1perm is the significance level that corresponds to the threshold for genome-wide significance. 

To control the confidence intervals, I would use drop=1.5 for a 1.5-LOD support interval (also see the function lod_int). The exact coverage unfortunately varies. You can also use prob=0.9 or prob=0.95 to get approximate Bayesian credible intervals (also see the function bayes_int). 

Also see the help file for find_peaks and the help file for summary.scan1perm.

karl