Chunk #43 — Statistical Methodology for the Analysis of Association Trajectories — Identification of significant SNPs — Significant SNPs identified by concentration of effect sizes
Any selection of SNP-age-phenotype combinations by thresholding of permutation scores will result in effect sizes specific to individual phenotypes which are less than 0.1 (or comparable value) being included in the select group, and effect sizes which are greater than 0.1 excluded from the select group. The exclusions are justified on the basis that they represent either sporadic occurrences over age and phenotype and therefore not of biological interest or else false positives; the inclusions reflect genetic heterogeneity in scalp localization in which large effect sizes in one region will counterbalance smaller effect sizes in other regions. Determination of the significance level for the p-values associated with the SNP-age combinations from the permutation was made by adapting the method of Storey and Tibshirani (2003) to determine the q-value (false discovery rate associated with the particular p-value), and establish the false discovery rate corresponding, if somewhat indirectly, to setting a threshold for effect sizes.
