For the genome-wide typed samples, duplicate samples, samples with a sex determined by X-chromosome homozygosity different from their reported sex and samples determined to be of non-European ancestry either by running STRUCTURE10 using the HapMap CEU, YRI and CHB/JPT individuals as reference samples or by examination of identity by state were removed. In addition, low-yield ( < 98% for the Illumina chips and <95% for the Affymetrix chip) samples were removed for the SNP association part of the study, whereas samples with an excess of large CNVs ( > 10 CNVs of at least 10 SNPs) were removed for the CNV part of the study. For the single-marker typed samples, samples that, based on additional genotyping, were duplicates of other samples in the data set or were low yield ( < 60%) were removed.
