LD patterns in dense variation data reveal information about the history of human populations worldwide. S. Myers1,2, G. Hellenthal2, D. Lawson3, G. Busby4, S. Leslie5, B. Winney5, P. Donnelly2, W. Bodmer5, The. POBI Consortium1,2,5, C. Capelli4, D. Falush6 1) Dept of Statistics, Oxford University, United Kingdom; 2) Wellcome Trust Centre for Human Genetics, Oxford University, United Kingdom; 3) Department of Mathematics, University of Bristol, United Kingdom; 4) Dept of Zoology, Oxford University, United Kingdom; 5) Cancer and Immunogenetics Group, Department of Clinical Pharmocology, Oxford University, United Kingdom; 6) Max Planck Institute for Evolutionary Anthropology, Department of Evolutionary Genetics, Leipzig.
"A detailed understanding of population structure in genetic data is vital in many applications, including population genetic analyses and disease gene mapping, and relates directly to human history. However, there are still few methods that directly utilize information contained in the haplotypic structure of modern dense, genome-wide variation datasets. We have developed a set of new approaches, founded on a model first introduced by Li and Stephens, which fully use this powerful information, and are able to identify the underlying structure in large datasets sampling 50 or more populations. Our methods utilize both Bayesian model-based clustering and principal component analyses, and by using LD information effectively, consistently outperform existing approaches in both simulated and real data. This allows us to infer ancestry with unprecedented geographical precision, in turn enabling us to characterize the populations involved in ancient admixture events and, critically, to precisely date such events. We applied our new techniques to combined data for 30 European populations sampled by us, or publicly available, and the worldwide HGDP data. We find almost all human populations have been influenced by mixture with other groups, with the Bantu expansion, the Mongol empire and the Arab slave trade leaving particularly widespread genetic signatures, and many more local events, for example North African (Moroccan) admixture into the Spanish that we date to 834-1394AD. Dates of admixture events between European groups and groups from North Africa and the Middle East, seen in multiple Mediterranean countries, vary between 800 and 1700 years ago, while Greece, Croatia and other Balkan states show signals of admixture consistent with Slavic migration from the north, which we date to 600-1000AD. At the finest scale, we are able to study admixture patterns in data gathered by a project (POBI) examining people within the British Isles. Our approaches reveal genetic differences between individuals from different UK counties, and show that the current UK genetic landscape was formed by a series of events in the millennium following the fall of the Roman Empire."
Bookmarks