SNPs and indels with an imputation quality score <0.3 (as defined by the RSQR_HAT value in MACH) or a minor allele frequency (MAF) <0.01 were excluded. Primary association analysis was performed separately within each platform family (HumanHap, OmniExpress and Affymetrix). For imputed SNPs, the estimated number of effect alleles (ranging from 0 to 2) was used as a covariate. For BMI, we conducted linear regression adjusting for study (indicator variables including cohort as well as primary GWAS outcome), age at blood draw and the top four principal components. For VTE, we conducted logistic regression adjusting for study as above and the top four principal components. For both BMI and VTE, we combined platform family-specific results with fixed-effects meta-analysis using the METAL [52] software. We used the Cochran’s Q statistic to test for heterogeneity across studies.
