the ARIC cohort18,30 and for the QIMR cohort14,29. We then estimated pairwise genetic relationships between individuals14 and removed one of each pair of individuals with an estimated relatedness of >0.025. After these quality control steps, 6,654 and 3,924 unrelated individuals were retained in the ARIC and QIMR cohorts, respectively. All the ARIC samples were from adults and the QIMR samples were from 3,247 adults and 677 16-year-old adolescents. The SNP data for both ARIC and QIMR cohorts were imputed to the HapMap Phase 2 CEU panel by MACH31. We used the best guess genotypes of the imputed SNPs and excluded imputed SNPs with HWE P value of <1 × 10–6, imputation R2 of <0.3 or MAF of <0.01 and retained 2,406,652 and 2,410,957 SNPs in the ARIC and QIMR cohorts, respectively. The ARIC cohort is part of the discovery sample of the GIANT meta-analysis, whereas the QIMR cohort is not. In the prediction analyses, the height and BMI phenotypes in the ARIC and QIMR cohorts were adjusted for age and sex effects and standardized to z scores14,18. In the QIMR cohort, only samples from 3,247 adults were used in the prediction analysis for BMI.
