Prior to meta-analysis, SNPs with a minor allele frequency (MAF) < 0.01 and poorly imputed SNPs (proper_info ≤ 0.4 (SNPTEST); r2hat ≤ 0.3 (mach2qtl)) were filtered. Genomic control (GC)29 was applied to adjust the statistics generated within each cohort (see Supplementary Table 1 for individual study λ values). Inverse variance fixed-effects meta-analyses were undertaken using different software packages METAL (2009-10-10 release)30 and GWAMA (version 2.0.6)31 by two meta-analysts in parallel and compared to obtain identical results. The meta-analysis results were obtained for a total of 2,684,393 SNPs. We applied a second GC correction to adjust the overall meta-analysis statistics (λ = 1.051) before selecting 21 SNPs for follow-up, which surpassed a P-value threshold of P < 1×10−5. This additional GC correction was, however, only applied for the purpose of choosing the arbitrary significance threshold; we report here Stage 1 P-values after only the first GC correction (see Supplementary Table 2) because a second GC correction is generally considered to be over-conservative32. We additionally selected SNP rs6537307 (P=4.3×10−5), which is in linkage disequilibrium with a known HHIP height-associated variant (HapMap r2
