286T Poster - Population Genetics
Thursday June 09, 8:30 PM - 9:15 PM
FSTruct: An FST-based tool for quantifying ancestry variability
Authors: Maike Morrison 1; Nicolas Alcala 2; Noah Rosenberg 1
Affiliations: 1) Stanford University; 2) International Agency for Research on Cancer
Keywords: Theory & Method Development
How variable are the ancestry vectors in a population structure plot? In other words, how much variation in estimated membership coefficient vectors exists across individuals in a population? Population structure inference methods such as STRUCTURE and ADMIXTURE are among the most widely used tools in modern population genetics, yet the variability of a population’s estimated ancestry is often determined only visually. We present a new method and R package, FSTruct, which computes a normalized variability statistic that equals 0 when the population’s ancestry is minimally variable and 1 when it is maximally variable. This property means that the method can be used to compare the ancestry variabilities of multiple populations, even populations with different numbers of individuals or substantially different mean ancestry vectors. The method is based on the population-genetic statistic FST, relying on an analogy between the variability of ancestry vectors across individuals and the variability of allele frequency vectors across populations. We demonstrate the method in a theoretical model of cluster memberships and in comparisons involving admixed populations, ancient genomics, and multiple data types, and we also introduce a bootstrap test for equivalence of two or more populations in their levels of ancestry variability. The R package is available on GitHub at github.com/MaikeMorrison/FSTruct.