We then adjusted for multiple comparisons for the number of polymorphic CNVs within each racial group by designing a permutation test. In this permutation, we randomly assigned deletions and their accompanied average admixture score to all AA subjects and computed the minimum p-value for each replicate set. We repeated this procedure 10,000 times to establish the genome-wide significance level with a family-wise error rate of 5%; and the nominal threshold was set at 3.50, which is the 5th percentile -log10(p) maxima over all 10,000 replicates. A test result was considered significant if it was lower than this threshold. Besides the PennCNV analysis, we used R to carry out our association analyses for this study (http://cran.r-project.org/).
