Likelihood-based inference of population history from low-coverage de novo genome assemblies

摘要: Short-read sequencing technologies have in principle made it feasible to draw detailed inferences about the recent history of any organism. In practice, however, this remains challenging due difficulty genome assembly most organisms and lack statistical methods powerful enough discriminate between recent, nonequilibrium histories. We address both inference challenges. develop a bioinformatic pipeline for generating outgroup-rooted alignments orthologous sequence blocks from de novo low-coverage short-read data small number genomes, show how such can be used fit explicit models population divergence admixture likelihood framework. To illustrate our approach, we reconstruct Pleistocene an oak-feeding insect (the oak gallwasp Biorhiza pallida), which, common with many other taxa, was restricted during ice ages longitudinal series southern refugia spanning Western Palaearctic. Our analysis sampled single each three major glacial reveals support unexpected dominated by admixture. Despite fact that 80% is affected last cycle, are able infer deeper these populations. These robust variation block length, mutation model sampling location individual genomes within refugia. This combination numerical calculation provides framework estimating applied organism without need prior genetic resources.