Significant SNPs and significant sex differences for the sex-modality-age combinations were determined by setting threshold effect sizes and p-values using a coordination of control of the false discovery rate by the method in Storey and Tibshirani (2003) using p-values (See Section 8.3.1 for details), and permutation tests on the absolute values of the effect sizes to establish SNP specificity of results (See Section 8.3.2 for details). Analysis of the distribution of p-values establishes a false discovery rate of less than 5% for a threshold of p < 10−4. Individual effect sizes were aggregated over age ranges and the nine electrode array in order to minimize effects of spurious large effect sizes on the identification of significant SNPs.
