1 Natural History Museum of Denmark, Faculty of Science, Københavns Universitet2 Computational and RNA Biology, Department of Biology, Faculty of Science, Københavns Universitet3 University of California4 Natural History Museum of Denmark, Natural History Museum of Denmark, Faculty of Science, Københavns Universitet5 Natural History Museum of Denmark, Natural History Museum of Denmark, Faculty of Science, Københavns Universitet6 Computational and RNA Biology, Department of Biology, Faculty of Science, Københavns Universitet
Over the last few years, new high-throughput DNA sequencing technologies have dramatically increased speed and reduced sequencing costs. However, the use of these sequencing technologies is often challenged by errors and biases associated with the bioinformatical methods used for analyzing the data. In particular, the use of naÃ¯ve methods to identify polymorphic sites and infer genotypes can inflate downstream analyses. Recently, explicit modeling of genotype probability distributions has been proposed as a method for taking genotype call uncertainty into account. Based on this idea, we propose a novel method for quantifying population genetic differentiation from next-generation sequencing data. In addition, we present a strategy to investigate population structure via Principal Components Analysis. Through extensive simulations, we compare the new method herein proposed to approaches based on genotype calling and demonstrate a marked improvement in estimation accuracy for a wide range of conditions. We apply the method to a large-scale genomic data set of domesticated and wild silkworms sequenced at low coverage. We find that we can infer the fine-scale genetic structure of the sampled individuals, suggesting that employing this new method is useful for investigating the genetic relationships of populations sampled at low coverage.