permutation replicates

[Saikat Bandyopadhyay]

What is a good number of permutation replicates when using the scan1perm() to get permutation-based significance thresholds? 

The documation suggests the n_perm is to be set on 1000. However, even with 30 cores, it takes a long time to run and I need to do this scan for more than 80 phenotypes in non-transformed and transformed serise.

I was wondering if the n_perm can be reduced, and if yes, to what extent?

[Karl Broman]

It depends how you intend to use the permutations. If you're trying to get an estimate of 5% significance threshold adjusting for the genome scan (the 95th percentile of the genome-wide maximum LOD in permutations), 1000 is generally required to get a good estimate. But the threshold isn't likely to vary a great deal between traits, so you could do the 1000 permutations for one trait and then use that as an approximate guide for the other traits. For traits that show little evidence of a QTL, you might just skip the permutations, which for traits that you can about that show an apparent QTL is could be worth the computation time.

karl