Assessing population differentiation and isolation from single‐nucleotide polymorphism data

摘要: Summary. We introduce a new, hierarchical, model for single-nucleotide polymorphism allele frequencies in structured population, which is naturally fitted via Markov chain Monte Carlo methods. There one parameter each closely analogous to population-specific version of Wright's FST, can be interpreted as measuring how isolated the relevant population has been. Our includes effects ascertainment and motivated by genetics considerations, explicitly transient setting after divergence populations, rather than equilibrium stochastic model, traditionally case. For sizes data set that we consider method provides good estimates considerably outperforms estimation methods those currently used practice. apply new existing human set, with different characteristics—the first consisting three close European populations; second four populations taken from across globe. A novelty our framework fit underlying assessed easily, these results are encouraging both sets analysed. analysis suggests Iceland more differentiated other two (France Utah), finding consistent historical record, but not obvious comparisons simple summary statistics.