Figure 2 provides a plot of the first two PCs from the EIGENSTRAT analysis. The first PC is a dimension anchored at one end by individuals of self-reported East Asian ancestry and at the other end by individuals of self-reported European ancestry. The second PC is a dimension that differentiates individuals of self-reported African ancestry from those of European ancestry. Additional PCs tended to distinguish other groups such as the Native Americans or those from India. To identify a core sample of individuals with European ancestry for genetic analysis, we computed the Mahalanobis D2 for each genotyped individual from the centroid of self-reported whites using all 10 PCs. Figure 1 displays the distribution of the resulting D2 values. The cluster of data points centered at a D2 value of approximately 3400 represents individuals who self-reported as Asian. We defined the European cluster as including all those individuals with D2 < 84. We also confirmed that full siblings and DZ twins were never in separate clusters. After adding 1087 MZ cotwins, the final sample of 7702 individuals with European ancestry (91.6%
