Chunk #39 — Statistical Methodology for the Analysis of Association Trajectories — Identification of significant SNPs — Significant SNPs identified by concentration of effect sizes
pseudo-SNPs. That is, 9 distinct permutations of the effect sizes of the 21 SNPs are produced, one for each electrode, resulting in a matrix of 9 × 21 values for each sex-modality-age combination. Then the effect sizes are averaged across electrodes for each of the 21 pseudo-SNPs created by the permutation process. This is done for each sex-modality combination, resulting in 4 averages from each of the seven age ranges for each pseudo-SNP for the observed data. Randomly permuting the SNPs contributing to the averaging across the electrodes ensures that the effect sizes from all the SNPs are used. Each permutation procedure instance produces the same number of random groups of values for pseudo-SNPs as is produced for the actual SNPs. No permutations of age or phenotype are performed in order to preserve the covariance of those items from the original calculation. A large number (1000) of permutation sets were calculated, and for each of the seven age ranges, the mean and standard deviations over all the pseudo-SNPs in each of the permutation procedures are calculated to provide the statistical properties associated with the null hypothesis. It should be noted that the permutation test results for any one SNP-age-sex-phenotype combination,
