Bayesian inference with Mr Bayes Molecular Phylogenetics exercise
Bayesian inference with Mr. Bayes Molecular Phylogenetics – exercise Ben Stöver WS 2014/2015
Bayesian inference with Mr. Bayes 1. Mr. Bayes 2 1. 1 Overview on Mr. Bayes • • • Implements the Metropolis-Hastings-Green-algorithm. Analyses can be started from the internal command line (just like in PAUP*). The help command displays a list of available commands. help <command> displays help to a certain commend. More information under http: //mrbayes. sourceforge. net/mb 3. 2_manual. pdf Command reference under http: //mrbayes. sourceforge. net/Help/ Ben Stöver 12. 02. 2015 Winter term Fortgeschrittenenmodul Molecular
Bayesian inference with Mr. Bayes 1. Mr. Bayes 3 1. 2 Important commands to start an analysis • execute <alignment. File>: Reads alignment data from a file (e. g. Nexus). • lset: Can be used to set the model parameters (e. g. nst=6 rates=invgamma sets the GTR+I+G model) • mcmc: Starts an MCMC analysis with the loaded alignment. • ngen: The number of generations for the chain to run. • samplefreq: The frequency of generations when a tree shall be written to the log file • printfreq: The frequency to print information on the current run to the command line. • diagnfreq: The frequency to analize the current process (e. g. the measure of the similarity of the current tree samples). • savebrlens: Specify yes here, if branch lengths shall be saved. • nchains: Specifies the number of parallel chains (default is 4). Ben Stöver 12. 02. 2015 Winter term Fortgeschrittenenmodul Molecular
Bayesian inference with Mr. Bayes 1. Mr. Bayes 4 1. 3 Important commands to finish an analysis • plot: time (The • sump: manual) • sumt: Ben Stöver Plots the deveopment of an estimated parameter over parameter to be displayed can be selected. ) Summarize samples of model parameters (section 2. 2. 8 in Summarize the tree samples 12. 02. 2015 Winter term Fortgeschrittenenmodul Molecular
Bayesian inference with Mr. Bayes 1. Mr. Bayes 5 1. 4 Running an analysis • During an analysis the current ln(linkelihood) of all chains are printed out. run 1 diagnosis output heated chains Ben Stöver 12. 02. 2015 run 2 cold chain Winter term Fortgeschrittenenmodul Molecular
Bayesian inference with Mr. Bayes 1. Mr. Bayes 6 1. 5 Continuing the analysis • After the specified number of generations has been reached, you are asked, if you want to continue the chain. • The average deviation of split frequencies should be below 0. 01 in order to stop the chain Ben Stöver 12. 02. 2015 Winter term Fortgeschrittenenmodul Molecular
Bayesian inference with Mr. Bayes 1. Mr. Bayes 7 1. 6 Output files • Mr. Bayes produces several output files in your working directory • run 1. p and run 2. p contain the development of the model parameters during the chain • Can be opened e. g. in Excel • run 1. t and run 2. t contain the tree samples collected during the chain • In NEXUS format • Can be opened with Tree. Graph 2 (or other tree editors) • After calling sumt, the result tree is contained in the con-file. Ben Stöver 12. 02. 2015 Winter term Fortgeschrittenenmodul Molecular
- Slides: 7