Principal component analysis
The PCA for genomic variation in the 191 Swedish accessions in Figure 2E was performed using a compressed mixed linear model run in R implementing the GAPIT package (Lipka et al., 2012; Zhang et al., 2010). Genotype data was downloaded from the 1001 genomes dataset filtered for bialleleic SNPs with >5% MAF (minor allele frequency). The first two PCs were used to split the accessions into the categories in Figure 2D and F.