Pairwise relatedness coefficients (π^) were estimated with plink using a LD pruned and MAF filtered set of SNPs (snps-only, window size=5,000, step size=300, r2<0.05, MAF>0.05). Principal Component Analysis was conducted using the same set of LD pruned and MAF filtered SNPs, with random removal of one member of each pair with a relatedness coefficient (π^) higher than 0.2. Eigenvectors were inferred using EIGENSOFT version 6.1.4105 on the relatedness-pruned set of individuals, and subsequently projecting all individuals onto those eigenvectors based on their genotypes. Individuals with all four grandparents born in Denmark were used as a reference for constructing a 3-dimensional ellipsoid using principal components 1, 2 and 3 with a radius of 5 standard deviations from the mean. Individuals located outside this ellipsoid were removed prior to the testing for batch effects. For each of the batch variables, each genotyped marker was tested for association with each batch versus the remaining batches pooled, using plink v1.90b4.
