Statistical analyses were performed at Texas Biomedical Research Institute using SOLAR (Blangero & Almasy 1998), except where noted. Genome-wide analyses in SOLAR use measured genotype association, which applies a likelihood ratio test to an additive model of allelic effect while including a covariance matrix of pairwise relatedness to control for kinship. Maximum drinks was inverse normalized to a mean of zero and a standard deviation of 3 prior to analysis. As a continuous trait, maximum drinks (N = 272 males, h2 = 0.49, p = 0.002) is expected to have greater statistical power than the categorical AD trait (Almasy & Blangero 2001), with 80% power to detect variants explaining at least 13.3% of the variance. The case-control AD analysis of 102 male cases and 212 male controls (h2 = 0.46, p = 0.018) spread across 81 pedigrees has 80% power to detect variants with a genotype relative risk of 2.5. The analysis of flushing response (108 male cases, 190 male controls h2 = 0.47, p = 0.038) has 80% power to detect variants with a genotype relative risk of 2.4. Age was included as a covariate for all analyses.
