Estimating time-varying selection coefficients from time series data of allele frequencies

Abstract

7 Time series data of allele frequencies are a powerful resource for detecting and 8 classifying natural and artificial selection. Ancient DNA now allows us to observe 9 these trajectories in natural populations of long-lived species such as humans. Here, 10 we develop a hidden Markov model to infer selection coefficients that vary over time. 11 We show through simulations that our approach can accurately estimate both selection 12 coefficients and the timing of changes in selection. Finally, we analyze some of the 13 strongest signals of selection in the human genome using ancient DNA. We show that 14 the European lactase persistence mutation was selected over the past 5,000 years with 15 a selection coefficient of 2-2.5% in Britain, Central Europe and Iberia, but not Italy. In 16 northern East Asia, selection at the ADH1B locus associated with alcohol metabolism 17 intensified around 4,000 years ago, approximately coinciding with the introduction of 18 rice-based agriculture. Finally, a derived allele at the FADS locus was selected in 19 parallel in both Europe and East Asia, as previously hypothesized. Our approach is 20 broadly applicable to both natural and experimental evolution data and shows how 21 time series data can be used to resolve fine-scale details of selection. Page 1 bioRxiv preprint doi: https://doi.org/10.1101/2020.11.17.387761; this version posted November 18, 2020. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY 4.0 International license. 22