All the analyses described below were conducted using the relatedness and ancestry pruned dataset, but some analyses were in addition also performed among ancestry outliers (see below, Supplementary Figures S24-S28 and Supplementary Tables S23-S28). A second PCA was performed on the sample excluding ancestry outliers, using EIGENSOFT105 as implemented in the Ricopili-pipeline104, which was used for adjusting for any remaining population substructure in downstream analyses.
