Bayesian Variable Selection for Detecting Adaptive Genomic Differences Among Populations

摘要: We extend an Fst-based Bayesian hierarchical model, implemented via Markov chain Monte Carlo, for the detection of loci that might be subject to positive selection. This model divides Fst-influencing factors into locus-specific effects, population-specific and effects are specific locus in combination with population. introduce a auxiliary variable each effect automatically select nonneutral effects. As by-product, efficiency original approach is improved by using reparameterization model. The statistical power extended algorithm assessed simulated data sets from Wright–Fisher migration. find inclusion selection suggests clear improvement discrimination as measured area under receiver operating characteristic (ROC) curve. Additionally, we illustrate discuss quality newly developed method on basis allozyme set fruit fly Drosophila melanogaster sequence wild tomato Solanum chilense. For small sample sizes, high mutation rates, and/or long sequences, however, methods based nucleotide statistics should preferred.