To conduct SNP-by-SNP genome-wide association analysis of BMI, we started initially with919,509 autosomal markers with a genotyping call rate greater than 95%. Further steps involved removing SNP markers that show evidence of deviation from Hardy-Weinberg Equilibrium (HWE) in 492 unrelated, self-identified white individuals extracted from the entire sample. These 492 individuals were selected via a two-step process. First, we focused on the homogenous self-identified white sample followed by the random selection of one individual from each biological relationship pair. In all, 6,237 autosomal SNPs were flagged for potential deviation from HWE (p < 0.001) and removed from the genome-wide association analysis. On the basis of minor allele frequency (MAF), we further removed SNP markers with an MAF < 0.01 (32,313). Therefore, the final genome-wide association marker set includes 880,959 autosomal SNPs.As noted previously, the sample of N=1,888 individuals with clean genotypes includes two MZ twin pairs. For the purposes of the genome-wide association analysis, we removed one individual (randomly) from each of the two MZ twins pair resulting in a final analysis sample of N=1,886.
