This preview shows pages 1–3. Sign up to view the full content.
Integrative Biology 200A
University of California, Berkeley
PRINCIPLES OF PHYLOGENETICS
Spring 2008
Lab 11: MrBayes Lab
Introduction
MrBayes uses a Markov Chain Monte Carlo (MCMC) approach to search for trees.
There
are two phases in an MCMC.
The first phase, called “burnin”, is more like a normal
heuristic search in which the program explores the parameter space trying to find parameter
values near the maximum likelihood.
During this phase the likelihood of the trees increases
steadily.
In the second phase, called stationarity, the program explores the parameter space
around the maximum likelihood.
During stationarity the likelihood seams to vary
stochastically around a mean value.
The cool thing about an MCMC is that when it is in stationarity it samples the parameter
values in a proportion that approximates the posterior probability.
That is to say, if given
your data and your model there is a 20% chance that Tree A is the right tree, then in
stationarity a MCMC will produce that tree approximately 20% of the time.
This is true not
only for trees, but for all the parameters.
Thus you can generate a distribution for the value
of any parameter.
It makes this approximation without holding any of the other parameters
constant, but instead integrating over all values of those parameters.
Of course this only
works if your model is correct, your priors are correct and you are actually in stationarity for
long enough.
It is not always easy to tell if you are in stationarity, or how long ‘long enough’
is.
Today we will explore Bayesian phylogenetics as discussed in lecture by running analyses and
interpreting results.
The goals of the lab are the following:
I.
Adding a MrBayes block to a nexus file
II.
Conduct a Bayesian analysis
III.
Analyze chain diagnostics and “burnin”
IV.
Summarize parameters from the posterior probability distribution
V.
Generate a tree with posterior probability values
MrBayes
is available free in Mac, PC or Unix from: http://mrbayes.csit.fsu.edu/index.php
The program comes with example data that we’ll use for the following exercises.
We’ll use
PAUP or a text editor to edit data matrices and EXCEL to analyze results from the Bayesian
analysis.
I. Adding a MrBayes block to a nexus file
A MrBayes block is a string of commands at the end of a data matrix file that tells the
MrBayes program what to do.
In the MrBayes block you can specify the model you want to
use, your prior distributions, the length of the analysis, the number of samples to keep, etc.
1
This preview has intentionally blurred sections. Sign up to view the full version.
View Full DocumentEach line must end with a “
;
”.
Once written, you simply have to execute your matrix in the
MrBayes program and it will do the rest.
The alternative is to execute your data matrix in
MrBayes and enter the commands there.
The commands are the same no matter where you
enter them.
1.
This is the end of the preview. Sign up
to
access the rest of the document.
 Spring '08
 LINDBERG,MISHLER,WILL

Click to edit the document details