SNP–BMI association at the <i>FTO</i> locus in multiple ethnic groups.

In each group, two separate panels were characterized and combined by meta-analysis: Amagasaki and Fukuoka panels in Japanese, Hanoi and Thai Binh panels in Vietnamese, and Ragama and estate panels in Sri Lankans (see Methods). BETA and SEM of each trait are shown as z-score after adjustment for standard deviation. Besides the SNPs listed in Table 2, two SNPs—rs62033408 and rs72805612 (in complete LD with rs7188250)—were tested for BMI association, because these showed significant association in African-Americans [7]. Phetero; P-value for Inter-population heterogeneity among three Asian groups. rs9939609 is in complete LD (r2 = 1.000) with rs62033400 in both HapMap CEU and HapMap JPT+CHB, whereas they are not in strong LD (r2 = 0.065) in HapMap YRI. In African Americans, rs62033400 showed strong association with BMI (P = 1.1E-5). In HapMap JPT+CHB, rs9923544 and rs9941349 are in complete LD (r2 = 1.00). Results for Europeans are drawn from the GIANT consortium (http://www.broadinstitute.org/collaboration/giant/index.php/GIANT_consortium) and those for African Americans are from the published data [7]; the effect allele frequencies in Europeans are from HapMap CEU data. *In European populations, the results for a proxy, rs1558902, which is in complete LD (r2 = 1.0) with rs1421085 in HapMap CEU, were reported.