For the permutation procedure, Table I gives the estimated asymptote , half–saturation parameter and genomewide significance threshold for the NBS and 58BC samples separately and combined. It is clear that the estimates are similar for the separate cohorts, so they may be combined to give greater precision. Figure 1(a) shows the threshold for 5% family–wise error plotted as a function of marker density for the combined samples. Figure 1(b) shows the corresponding effective numbers of tests together with the fitted Monod function. The curve is a good fit, but it is clear that it is not at its asymptote at the current density, although some curvature is apparent when compared with the linear regression line. The estimated asymptote was = 651,550, which increases to 693,138 assuming that the autosomes comprise 94% of the total genome length. This gives our estimated genomewide significance threshold as (11) with 95% bootstrap confidence interval (6.3–8.9) × 10−8.
