Inference of Selection from Genetic Time Series Using Various Parametric Approximations to the Wright-Fisher Model

Abstract

Detecting genomic regions under selection is an important objective of population genetics. Typical analyses for this goal are based on exploiting genetic diversity patterns in present time data but rapid advances in DNA sequencing have increased the availability of time series genomic data. A common approach to analyze such data is to model the temporal evolution of an allele frequency as a Markov chain. Based on this principle, several methods have been proposed to infer selection intensity. One of their differences lies in how they model the transition probabilities of the Markov chain. Using the Wright-Fisher model is a natural choice but its computational cost is prohibitive for large population sizes so approximations to this model based on parametric distributions have been proposed. Here, we compared the performance of some of these approximations with respect to their power to detect selection and their estimation of the selection coefficient. We developped a new generic Hidden Markov Model likelihood calculator and applied it on genetic time series simulated under various evolutionary scenarios. The Beta with spikes approximation, which combines discrete fixation probabilities with a continuous Beta distribution, was found to perform consistently better than the others. This distribution provides an almost perfect fit to the Wright-Fisher model in terms of selection inference, for a computational cost that does not increase with population size. We further evaluated this model for population sizes not accessible to the Wright-Fisher model and illustrated its performance on a dataset of two divergently selected chicken populations.

Gene Banks as Reservoirs to Detect Recent Selection: The Example of the Asturiana de los Valles Bovine Breed

“…The present study, framed within the IMAGE project, aimed at enhancing the use of gene bank collections in animal breeding. To outline the invaluable resources provided by periodic resampling and cryopreservation of germplasm and tissue samples from local less competitive populations, we analyzed existing and new data from the Spanish Asturiana de los Valles beef cattle breed, using a new statistical approach that takes advantage of genomic time series to detect and characterize recent selection signatures in a population ( Paris et al, 2019 ). Results from such selection scans will help the livestock industry to exploit the genetic variation fine-tuned by natural and human-driven selection on different breeds to improve productivity.…”

“…We detected the loci that have been under selection in the Asturiana de los Valles breed between 1980 and 2013 using a new method that exploits the evolution of allele frequencies in a population along different sampling times ( Paris et al, 2019 ). This method is based on a HMM approach, which allows for the modeling of both the stochastic evolution of population allele frequencies over time, as a result of genetic drift and selection (if any), and the additional noise arising from the finite sample size at each time point.…”

“…This method is based on a HMM approach, which allows for the modeling of both the stochastic evolution of population allele frequencies over time, as a result of genetic drift and selection (if any), and the additional noise arising from the finite sample size at each time point. Other similar HMM approaches were previously proposed in the literature, but they were either less accurate or limited for computational reasons to very small population sizes; see Paris et al (2019) for more details. We applied the time series approach either with or without WGS samples (the number of individuals and SNPs for the two analyzed are summarized in Table 1 ); for SNPs that were shared by the two datasets, we kept the p -value computed with WGS samples, as this corresponds to the larger sample.…”

“…SNPs with a MAF below 5% over all the sampling times were also removed from this time series analysis. Indeed, some assumptions of the Likelihood Ratio Test used by Paris et al (2019) to detect selection are not satisfied for rare alleles, which may lead to less accurate p -values. Besides, such SNPs correspond to allele frequency trajectories showing little variation over time, which are unlikely to contribute to significant evidence of selection in a time series analysis.…”

“…To illustrate this approach and detect selection using either temporal or present time sampling, we combined SNP chip data from previous projects and whole-genome sequencing (WGS) data for the Asturiana de los Valles breed produced within the European Project IMAGE, and built a dataset covering eight generations of this population. These data were used to detect selection signatures in this breed using both a new statistical approach taking advantage of genomic time series ( Paris et al, 2019 ) and a state of the art statistic (nSL) based on a single sampling time ( Ferrer-Admetlla et al, 2014 ). Apart from expanding our understanding on the genomic grounds of Asturiana de los Valles evolution and providing molecular tools for enhancing the performance of this breed, this study outlines one potential use of gene bank collections in animal breeding.…”