NSF Award
1456634
The ability to identify the geographic origin of an individual using genomic data poses a great challenge due to its complexity and potential misinterpretations. Knowledge of this origin and recent ancestry are essential for research in multiple fields such as anthropology, sociology, forensics, personalized medicine and epidemiology, in which ancestry is an important variable. It also requires understanding that all species, including humans, are mixed to certain degrees and that these mixture patterns can unlock the history and origin of their ancestors. As the proportion of mixed-ancestry individuals increases worldwide, there is a need to better infer their biogeography. Current methods are less than 50% accurate for European populations and highly inaccurate for non-Europeans. This project aims to address this shortage and to develop novel, accurate and efficient tools to study individuals of mixed origin. They will have vast implications for practitioners trying to match cases and controls in disease studies, geneticists studying biodiversity and origins of humans, animals, and plants, as well as many people seeking answers about their past. This work will also contribute to advancement in agricultural genomics by providing selection tools for plant and animal breeders.<br/><br/><br/>Recently published first-generation Geographic Population Structure (GPS1) algorithm, developed by the PIs, provided biogeographical predictions that placed 83% of worldwide non-admixed individuals in their correct country of origin. This proposal builds on the success of the GPS1 algorithm to develop new tools for predicting biogeography in mixed individuals. The current aims are: (1) Development of the next phase of GPS algorithms, which will be capable of predicting the countries of origin of an individual's parents, grandparents or a more complex mixture with high accuracy; (2) Development of a tool to infer local ancestry along the genome; (3) Development of a GPS pipeline to infer the biogeographic origin of plants and animals. Modern computational approaches, such as genetic algorithms, simulated annealing, and others will be used to achieve optimal accuracy and computational efficiency. All algorithms will be implemented in the platform-independent languages R and Matlab and use the mpiR R package and parallel computing toolbox, respectively, to enable parallel processing. This project is supported by the Evolutionary Processes and Biological Anthropology programs at NSF.
