Scribd
Upload
68 views2 pages

BEAUTi 1.5.1 Multi-Partition Calibration Guide

This document discusses the multi-partition structure and calibration options in BEAST version 1.5.1. It describes that BEAST allows for multiple substitution models, clock models, tree models, and tree priors to be applied to partitions of the data. It also outlines five categories of calibration that can be used: 1) estimating relative clock rates with a fixed mean, 2) estimating rates relative to a reference clock, 3) calibrating tip dates, 4) calibrating internal nodes, and 5) calibrating clock rates. It provides examples of proper prior distributions that can be used for calibration, including lognormal, normal, Laplace, truncated normal, gamma with shape >1, gamma with offset >0, uniform with bounds >

Uploaded by

kcdeitz7690
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
68 views2 pages

BEAUTi 1.5.1 Multi-Partition Calibration Guide

This document discusses the multi-partition structure and calibration options in BEAST version 1.5.1. It describes that BEAST allows for multiple substitution models, clock models, tree models, and tree priors to be applied to partitions of the data. It also outlines five categories of calibration that can be used: 1) estimating relative clock rates with a fixed mean, 2) estimating rates relative to a reference clock, 3) calibrating tip dates, 4) calibrating internal nodes, and 5) calibrating clock rates. It provides examples of proper prior distributions that can be used for calibration, including lognormal, normal, Laplace, truncated normal, gamma with shape >1, gamma with offset >0, uniform with bounds >

Uploaded by

kcdeitz7690
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

2 September 2009

BEAUTi 1.5.1
Multi-Partition Structure:
PartitionSubstitution Model 1...* 1 PartitionData 1 1...* PartitionClockModel

1 1...* PartitionTreeModel 1 1...*

1 1...* PartitionClockModel TreeModelLink

1 1 or * PartitionForSpecial Case 1 1 PartitionTreePrior

Given n PartitionData, it can create: i PartitionSubstitutionModel (1 i n), j PartitionClockModel (1 j n), k PartitionTreeModel (1 k n), l PartitionTreePrior (l = either 1 or k), m PartitionClockModelTreeModelLink (m = j k), and partition used for special case, where n, i, j, k, l, m positive Integer In BEAUTi 1.5, the value of i, j, k, can be changed in Data Partitions panel by using buttons Unlink Subst Models, Link Subst Models, Unlink Clock Models, Link Clock Models, Unlink Tree Models, and Link Tree Models respectively. l is controlled by check box Share the same tree prior in Tree panel. m is automatically calculated after each change of j or k.

But for *BEAST, the default setting is i = n, j = n, k = n, l = 1.

Page 1

2 September 2009

Five Categories Of Calibration:


There are five categories of calibration in BEAUTi 1.5: 1) Estimate relative clock rates (fixed mean) This only applies to multi-clock models, namely j > 1. In this scenario, the delta exchange operator deltaAllClockRates is used to estimate all clock rates, with the constraint that their mean is fixed to a specific user-specified value (1.0 by default). 2) Estimate clock rates relative to x partition clock model (where x {1, , j}) In this scenario, xth clock rate is fixed to a given value, and the rest of clock rates are estimated relative to this fixed rate. 3) Tip times calibrated This scenario is triggered, if the maximum tip height in Tip Dates panel is greater than 0. 4) Internal node(s) calibrated This scenario is triggered, if either there is a proper prior on parameter [Link], or there is a proper prior on tmrca (taxon set) statistic. 5) Clock rate(s) calibrated This scenario is triggered, if there is a proper prior on clock rates.

What Is A Proper Prior?


A proper prior is: 1) Lognormal 2) Normal 3) Laplace 4) Truncated Normal 5) Gamma AND shape > 1 6) Gamma AND offset > 0 7) Uniform AND lower > 0 AND upper < + 8) Exponential AND offset > 0

Page 2

You might also like