We excluded individuals whose putative geographic origin was from outside of Europe (for example, Europeans from USA, China, Mozambique, Ivory Coast, and so on), individuals who were putatively related (using the same approach as in ref. 7), and individuals found to be outliers in a preliminary PCA run (for more detail, see the section on PCA below). Because of the large number of Swiss individuals available and the availability of language information for most of these individuals, for some analyses, we divided Swiss individuals into three ancestry labels (Swiss-French, Swiss-German and Swiss-Italian) on the basis of their reported primary language. Finally, we chose to include only a random sample of 200 individuals from the United Kingdom and 125 Swiss-French to obtain more even sample sizes across Europe. Supplementary Table 2 provides more detail on how the sample numbers changed with each step in the sample preparation, and Supplementary Table 1 summarizes the number of grandparents observed for the 1,387 individuals used in the final sample.
