Only SNPs having a minor allele frequency of 5% or greater that were within 10 kb of PPARA, PPARD, PPARG, and PPARGC1A were considered in the analysis. The Genome-Wide Association Analyses with Family Data package (Chen and Yang, 2010) implementing a log additive model was used for analysis of AD. The generalized disequilibrium test (Chen et al., 2009), employing data from all discordant relative pairs, was used for analysis of the withdrawal criterion. To account for secular trends, sex and birth cohort defined by year of birth (<1930, 1930–1949, 1950–1969, ≥1970), were used as covariates. In regions of interest, imputed SNPs were analyzed to further evaluate the evidence for association. SNPs were imputed to 1000 Genomes (EUR, August 2010 release) using BEAGLE 3.3.1 (Browning and Browning, 2009) as described in Wang et al. (2013).
