NSF PostDoctoral Research Fellowship in Biology

这项行动为2008财政年度的NSF生物学博士后研究奖学金提供资金。该研究金支持一项研究和培训计划，题为“替代率变化的复杂模型和估计物种分歧时间的贝叶斯方法的比较”。希思这项研究的主办机构是加州大学伯克利分校和堪萨斯大学，赞助科学家分别是John Huelsenbeck博士和Mark保持器博士。在进化生物学探索的问题中，理解进化过程的时间和速度一直是核心。系统发生树（描绘物种关系的分支图）是理解广泛进化模式的重要工具，经常用于估计物种形成事件的时间。这些物种的分歧时间的准确估计是至关重要的理解历史地理，估计物种形成和灭绝的速度，并确定分子进化速度的变化的原因。本研究的重点是发展新的分子模型，用于系统发育分析和分歧时间估计。这项研究提供了一个通用的和灵活的框架，通过使用一个称为“狄利克雷过程之前”的模型，将更生物学上现实的模式变化的分子进化到统计系统发育的方法。新的模型将使用贝叶斯推断方法（估计统计模型参数的强大方法）在免费提供的开放源码软件程序中实施。本课程将利用复杂分子模型下的模拟数据集，对贝叶斯方法进行全面比较，以确定物种分歧的年代。培训目标包括加强计算技能和编程能力。此外，由于该项目整合了分子进化，统计学和计算机科学，它将导致与各级研究人员的跨学科合作的培养。

This action funds an NSF Postdoctoral Research Fellowship in Biology for FY 2008. The fellowship supports a research and training plan entitled "Complex models of substitution rate variation and a comparison of Bayesian methods for estimating species divergence times" for Tracy A. Heath. The host institution for this research is University of California-Berkeley and University of Kansas, and the sponsoring scientists are Dr. John Huelsenbeck and Dr. Mark Holder, respectively. Among the questions explored in evolutionary biology, understanding the timing and rates of evolutionary processes has been central. Phylogenetic tree (branching diagrams depicting species relationships) are essential tools for understanding broad evolutionary patterns and are frequently used to estimate the timing of speciation events. Accurate estimates of these species divergence times are vital to understanding historical biogeography, estimating rates of speciation and extinction, and identifying the causes of variation in rates of molecular evolution. This research is focusing on the development of new molecular models for phylogenetic analysis and divergence time estimation. The research provides a general and flexible framework for incorporating more biologically realistic patterns of variation in the rates of molecular evolution into statistical phylogenetic methods by using a model called the "Dirichlet process prior". New models will be implemented in freely available, open source software programs using Bayesian inference methods (powerful methods for estimating the parameters of a statistical model). Simulated data sets generated under complex molecular models will be used to conduct a thorough comparison of Bayesian methods for dating species divergences.Training objectives include strengthening computational skills and programming ability. Moreover, because this project integrates molecular evolution, statistics, and computer science, it will lead to the cultivation of interdisciplinary collaborations with researchers at all levels.