build 36 (Ensembl build 54, Hg18) using BLAT, BWA and SOAPv2 sequence alignment programs. Highly stringent alignment criteria were used to ensure that probes map unequivocally to one single genomic position. Genotype data was acquired using different genotyping platforms, and harmonized by imputation, using the HapMap247 Central European population as a reference. Each dataset was individually checked for sample mix-ups using MixupMapper48. For a full description of the individual datasets, results of the sample mix-up analysis, specifics on the gene expression platforms and probe mapping procedure and filtering, see Supplementary methods.
