第五章多序列对位排列和进化分析.ppt

最大简约法 (Maximum Parsimony) 最大简约法（MP）最早源于形态性状研究，现在已经推广到分子序列的进化分析中。最大简约法的理论基础是奥卡姆（Ockham）哲学原则，对所有可能的拓扑结构进行计算，找出所需替代数最小的那个拓扑结构，作为最优树。 G A A Sequence4 G G A Sequence3 C A T Sequence2 C G T Sequence1 1 2 3 Position MP法建树流程 If 1 and 2 are grouped a total of four changes are needed.  If 1 and 3 are grouped a total of five changes are needed.  If 1 and 4 are grouped a total of six changes are needed. Position 1 (1,2): 1 change;  (1,3) or (1,4): 2 changes Position 2 (1,3): 1 change;  (1,2) or (1,4): 2 changes Position 3 (1,2): 1 change;  (1,3) or (1,4): 2 changes 4 5 6 BEST MP法建树步骤 最大似然法 (Maximum Likelihood) 最大似然法(ML) 最早应用于对基因频率数据的分析上。其原理为选取一个特定的替代模型来分析给定的一组序列数据，使得获得的每一个拓扑结构的似然率都为最大值，然后再挑出其中似然率最大的拓扑结构作为最优树。 ML法建树流程 C A G ATGC C ATGC Pick an Evolutionary Model For each position, Generate all possible tree structures Based on the Evolutionary Model, calculate Likelihood of these Trees and Sum them to get the Column Likelihood for each OTU cluster. Calculate Tree Likelihood by multiplying the likelihood for each position Choose Tree with Greatest Likelihood Inferring the maximum likelihood tree HolderLewis (2003) Nature Reviews Genetics 4, 275-284 Bayesian inference: What is the probability that the model/theory is correct given the observed data? Pr(T|D) Maximum Likelihood: What is the probability of seeing the observed data (D) given a model/theory (T)? Pr(D|T) Speed No need for bootstrapping 构建进化树的新方法——贝叶斯推断(Bayesian inference) 与ML相比，BI的优势： Comparison of Methods Good for very small data sets and for testing trees built using other methods Best option when tractable (30 taxa, homoplasy rare) Good for generating tentative tree, or choosing among multiple trees Highly dependent on assumed evolution model Assumptions fail when evolution is rapid Easily trapped in local optima Very slow Slow Very fast Maximizes tree likelihood given specific parameter values Minimizes total