If there were autosomal data only or scanone was
run with perm.Xsp=FALSE, genome-wide LOD thresholds are given;
these are the 1-\(\alpha\) quantiles of the genome-wide maximum LOD
scores from the permutations.
If there were autosomal and X chromosome data and
scanone was run with perm.Xsp=TRUE,
autosome- and X-chromsome-specific LOD thresholds are given, by the
method described in Broman et al. (2006). Let \(L_A\) and
\(L_X\) be total the genetic lengths of the autosomes and X
chromosome, respectively, and let \(L_T = L_A + L_X\)
Then in place of \(\alpha\), we use
$$\alpha_A = 1 - (1-\alpha)^{L_A/L_T}$$
as the significance level for the autosomes and
$$\alpha_X = 1 - (1-\alpha)^{L_X/L_T}$$
as the significance level for the X chromosome. The result is a list
with two matrices, one for the autosomes and one for the X chromosome.
If controlAcrossCol=TRUE, we use a trick to control the error
rate not just across the genome but also across the LOD score
columns. Namely, we convert each column of permutation results to
ranks, and then for each permutation replicate we find the maximum
rank across the columns. We then find the appropriate quantile of the
maximized ranks, and then backtrack to the corresponding LOD score
within each of the columns.