Shanks, C. (BMEB) – Development and Application of Local Ancestry Methods for Population Genomics

Local ancestry methods classify the segments of DNA inherited from a specific ancestry (e.g., African, East Asian, European), improving analyses of admixed populations. Aim 1 applies ancestry-specific analysis to more than 1000 whole genome sequences across Polynesia, revealing strong bottlenecks in the voyagers who settled both Hawaiʻi and Rapa Nui, and confirming frequencies of Mendelian disease–causing variants in French Polynesia not found even in large-scale biobanks. Conventional local ancestry methods are only accurate for deeply diverged populations and recent admixture events. In aim 2 I present ARGMix, a new deep-learning approach incorporating ancestral recombination graphs (ARGs), which contains the inferred history of coalescent events with recombination. This method classifies the local ancestry of present day Europeans as originating from early European farmers and hunter gather ancestries. As an application I find evidence of continuity between Ötzi the Iceman and present-day Europeans of similar geography. I further propose to apply ARGMix to trace the TCC/TTC mutational pulse strongest in Europeans across geography and time. In aim 3 I present a method incorporating these ancient local ancestries to improve polygenic risk scores in the UK biobank.

Event Host: Cole Shanks, Ph.D Student, Biomolecular Engineering & Bioinformatics