Distributions of selection coefficients from phylogenomic data - modeling the time-heterogeneity of the substitution process

点击翻译按钮获取中文摘要

Thanks to the availability of a broad spectrum of bio-molecular data, and the capabilities of modern computing machines, the field of statistical modeling of molecular evolution has been undergoing rapid advances in recent years. Among these advances are a class of substitution models known as mutation-selection models, in which distinct parameters are assigned to mutational features and selective effects. These models are inspired by population-genetics theory, yet are applicable to the analysis of homologous genes sampled across several different species. Using the formalisms of probability theory, one can quantify the evidence for a given model, and by comparing the evidence for several alternative forms of mutation-selection models, one can understand the relative importance of the different elements bearing on the overall, long-term evolutionary process. Over the course of this research program, we will develop novel models of molecular evolution of protein-coding DNA sequences using the mutation-selection framework, which allow for different aspects of the evolutionary process to vary over time. These models entail non-analytical mathematical expressions, requiring the use of high-performance computing for obtaining numerical approximations of the desired probabilities. The program will enable a sound probabilistic exploration of fundamental questions in evolutionary genetics, regarding the relative importance of mutation, selection and drift, but also regarding how shifts of these features relate to shifts in species traits, population structures, and ecological features.