Evaluation of significance of in the RA candidategene study. For subsets of a given size, we plotted the highest GTD score (red solid line) with a 95% confidence interval (with Bonferroni correction for 220 - 1 multiple comparisons). The top GTD scores for sets of two to eight marker sets were significantly higher than the expected value under the null hypothesis (black dotted line at 1). Based on 100 permuted data sets, at different subset sizes, the black solid line displays the median maximum GTD scores and the vertical bars are the 95% confidence interval of permutations. For each permutation, we also calculated Bonferroni-corrected 95% confidence intervals for the maximum GTD scores. The blue shading indicates the coverage of these 100 confidence intervals at each subset size (the darkest being 0.9 to 1 (or 90 to 100%) and the lightest being 0.0 to 0.1 (or 0 to 10%)).