For maximal accuracy, we use reference populations that have been fully sequenced (complete genomes) rather than references that had been genotyped at only a subset of sites. These samples come from the 1000 Genomes research project, which sequenced full genomes from individuals around the world (1000 Genomes Project Consortium, 2012). For Native American references, we used samples within the 1000 Genomes project of Native American ancestry; these samples come from Mexico, Peru, and Colombia. (It is not possible to use Native American reference sequences from inside the United States, since Native American groups within the US have not chosen to participate in recent population genetics studies.) The 1000 Genomes reference samples come from Nigerian Yoruba individuals (for Sub-Saharan Africa), Finnish, Tuscan Italian, and Spanish individuals (for Europe), and northern Chinese individuals for East Asia. (The latter reference was used to test for East Asian regional ancestry, since that can otherwise be mis-assigned as Native American). In our analysis, an individual with 100% ancestry assigned to a single population (e.g., European or African) is defined as an “unadmixed”.