Evaluating the accuracy of AIM panels at quantifying genome ancestry
- Equal contributors
1 Unidade de Xenética, Departamento de Anatomía Patolóxica e Ciencias Forenses, and Instituto de Ciencias Forenses, Grupo de Medicina Xenómica (GMX), Facultade de Medicina, Universidade de Santiago de Compostela, 15872 Santiago de Compostela, Galicia, Spain
2 Grupo de Investigación en Genética, Vacunas, Infecciones y Pediatría (GENVIP), Hospital Clínico Universitario and Universidade de Santiago de Compostela (USC), Santiago de Compostela, Galicia, Spain
3 Unidad de Emergencias Pediátrica y Cuidados Intensivos, Departamento de Pediatría, Hospital Clínico Universitario de Santiago, Santiago de Compostela, Galicia, Spain
BMC Genomics 2014, 15:543 doi:10.1186/1471-2164-15-543Published: 30 June 2014
There is a growing interest among geneticists in developing panels of Ancestry Informative Markers (AIMs) aimed at measuring the biogeographical ancestry of individual genomes. The efficiency of these panels is commonly tested empirically by contrasting self-reported ancestry with the ancestry estimated from these panels.
Using SNP data from HapMap we carried out a simulation-based study aimed at measuring the effect of SNP coverage on the estimation of genome ancestry. For three of the main continental groups (Africans, East Asians, Europeans) ancestry was first estimated using the whole HapMap SNP database as a proxy for global genome ancestry; these estimates were subsequently compared to those obtained from pre-designed AIM panels. Panels that consider >400 AIMs capture genome ancestry reasonably well, while those containing a few dozen AIMs show a large variability in ancestry estimates. Curiously, 500-1,000 SNPs selected at random from the genome provide an unbiased estimate of genome ancestry and perform as well as any AIM panel of similar size. In simulated scenarios of population admixture, panels containing few AIMs also show important deficiencies to measure genome ancestry.
The results indicate that the ability to estimate genome ancestry is strongly dependent on the number of AIMs used, and not primarily on their individual informativeness. Caution should be taken when making individual (medical, forensic, or anthropological) inferences based on AIMs.