Like all biological populations, human groups can’t be neatly divided. Real populations are connected to each other, and their borders are blurred by migration and mixing. But when inferring ancestry of an individual from genetic data, populations are typically simplified into tidy, discrete units.
In the December issue of G3: Genes|Genomes|Genetics, Yang et al. describe a method for ancestry inference of admixed individuals that uses a geographic approach to explicitly model some of the messy realities of populations. Testing the new method on data from the Population Reference Sample project, the authors were able to localize the grandparents of admixed Europeans to within around 500 kilometres of their reported ancestry, while simultaneously identifying which segments of each person’s genome were inherited from each ancestor.
The method is built on a previous approach called spatial ancestry analysis (SPA), which models genetic variation as allele frequency gradients along a geographic continuum. The new method, SPAMIX, infers ancestry for admixed individuals (i.e., those with recent ancestors from several genetically diverged populations) by combining SPA with models of the admixture process.
Ancestry inference is not only useful for studying population histories; it’s also a critical tool in genome-wide association studies (GWAS) of human disease, where failing to control for differences in allele frequencies between population groups can generate spurious associations. The authors suggest their new method is not only applicable to human studies, but could prove useful for identifying the origins of recently hybridized organisms, including invasive species and pathogens.
Yang W.Y., C. W.-K. Chiang, E. Eskin, J. Novembre & B. Pasaniuc (2014). Spatial Localization of Recent Ancestors for Admixed Individuals, G3, 4 (12) 2505-2518. DOI: http://dx.doi.org/10.1534/g3.114.014274