Leveraging ancient DNA to elucidate the genetic architecture of complex human traits

About This Grant

Ancient DNA has transformed our understanding of demographic events in the past, but its potential to reveal as much about biology as history has not been realizable until recently because estimating allele frequencies with enough precision to detect significant shifts over time requires larger sample sizes than were previously available. In preliminary work, we analyzed whole-genome data from 8,433 West Eurasians dating to the last 14,000 years, a six-times larger dataset than any previous study of this type. To control for population structure, we took a novel approach of using a Generalized Linear Mixed Model, searching for a consistent trend in allele frequency change over all regions and times while allowing starting frequency to vary across subgroups. We find 347 independent loci passing our threshold for genome-wide significance, and many complex phenotypes shaped by directional selection, described in a preprint posted in September 2024 (Ali Akbari et al. bioRxiv). These developments highlight the promise of ancient DNA for providing insight into complex trait architecture, as for the first time we can directly measure the selection coefficients of large numbers of variants with known effects on traits from Genome-Wide Association Studies, making it possible to fit models of the relationship between the frequency distribution of causal alleles for diverse traits and their effects. To take advantage of the opportunities opened up by these technical and data breakthroughs, we propose a four-part work plan: (1) Assemble an ancient DNA time-transect of 34,000 West Eurasians by the end of the grant period, quadrupling sample size and allowing measurement of selection coefficients down to around s = 0.25%. (2) Curate the data into a publicly available resource, released pre-publication to accelerate science. (3) Analyze the data to learn about how selection forces over time shape the distribution of alleles modulating phenotypes, thus characterizing the “selection architecture” of diverse traits. (4) Carry out the first ancient DNA studies of selection in three time-transects outside West Eurasia, analyzing about 1000 individuals each spread over 10000 years in Brazil, the Andes, and Mesoamerica. The anticipated result is an advance in our understanding of how selection impacts humans, based on analyses in multiple populations in parallel thus enabling more generalizable conclusions that are maximally relevant to the largest number of people. Since selection is expected to be a primary shaper of the distribution of frequencies of alleles contributing to traits, this should provide distinctive insights complementing other approaches for investigating complex trait architecture.