DNA samples were submitted for genotyping under a number of primary projects using different Illumina SNP platforms (Human610-Quad, HumanCNV370-Quadv3 and Human 317K). Standard quality control (QC) filters were applied to the genotyping in the different platforms. QC included checks for ancestry outliers, Mendelian errors, Hardy Weinberg Equilibrium, and Minor Allele Frequency (MAF) and was conducted separately for each of the projects. Thereafter, the combined dataset was screened for missingness within individuals, pedigree and sex errors, and Mendelian errors. Full details of the initial QC procedures for the Illumina and Affymetrix data can be found elsewhere [24]. Imputation to the European reference dataset (HapMap 1+2, Release 22 Build 36) was undertaken by means of MACH [25] using a set of Single Nucleotide Polymorphisms (SNPSs) common across all genotyping platforms. SNPs characterized by either a low minor allele frequency (MAF<.01) or a low imputation quality score (R2<0.3) were removed. Monozygotic twins that were not genotyped were assigned their co-twin's genotype. The final dataset included ∼2.4 million imputed autosomal SNPs and 13,783 genotyped X-chromosomal SNPs available for association analysis.
