Tutorial 3.1
From BEAST Software
BEAST Documentation->Tutorials->Tutorial 3.1
As of BEAST v1.4.1, BEAUti can now be used to create taxon sets and add calibration priors
Calibrating the date of a MRCA using a bounded range or parametric probability distribution
First create a list of taxa that you wish to constrain the mrca of:
<taxa id="taxaSubset1"> <taxon idref="taxon_A"/> <taxon idref="taxon_B"/> <taxon idref="taxon_C"/> </taxa>
Then create a tmrcaStatistic that will record the height of the mrca of this taxa subset:
<tmrcaStatistic id="tmrca_subset1" name="tmrca_ABC"> <treeModel idref="treeModel"/> <mrca> <taxa idref="taxaSubset1"/> </mrca> </tmrcaStatistic>
Then create a uniform prior with the appropriate upper and lower limits:
<mcmc id="mcmc1" chainLength="100000" autoOptimize="true"> <posterior id="posterior"> <prior id="prior"> <uniformPrior lower="90" upper="100"> <tmrcaStatistic idref="tmrca_subset1"/> </uniformPrior > ... </prior> <likelihood id="likelihood"> <treeLikelihood idref="treeLikelihood"/> </likelihood> </posterior> ... </mcmc>
NOTE: it is very important that you ensure that the starting tree obeys any uniform prior constraints you might add (because the tree has zero prior probability outside these ranges). This can be achieved by defining a starting tree as explained in Tutorial 4. Also if you want the group of taxa to be monophyletic, then you must also use the code outlined in [Tutorial 5].
It is also possible to add other distributions as priors such as a normal distribution:
<mcmc id="mcmc1" chainLength="100000" autoOptimize="true"> <posterior id="posterior"> <prior id="prior"> <normalPrior mean="100" stdev="5"> <tmrcaStatistic idref="tmrca_subset1"/> </normalPrior> ... </prior> <likelihood id="likelihood"> <treeLikelihood idref="treeLikelihood"/> </likelihood> </posterior> ... </mcmc>
Exponential, Log Normal and Gamma distributions are also available as priors.
Alexei Drummond and Andrew Rambaut
Copyright © 2002-2006 All rights reserved.

