Of the combined 500K and 10K Mapping array sets, the association analyses focused on 362,129 SNPs that passed quality control checks (48, 49). The remaining SNPs failed quality checks (∼2.9% of SNPs failed checks for data completeness, Hardy-Weinberg equilibrium, and Mendelian incompatibilities) or had a minor allele frequency of <5% (∼25.7% of SNPs had low minor allele frequencies). Although the five factor scores are approximately normally distributed, to avoid inflated type I error rates an inverse normal transformation was applied to all phenotype variables prior to analysis. Association analyses were carried out as described elsewhere (8, 47). The additive genetic effect was estimated for each SNP in the context of a variance component model that accounts for resemblance among related individuals (47). We analyzed each of the five factor scores in turn, including sex, age, and age2 as covariates.
