Welcome! Here you can find information and download files related
the CoMET project.
What is CoMET?
CoMET stands for Continuous-character Model
Evaluation and Testing
and is a module for the
Mesquite Project. Give it a tree topology as a
starting point and some phenotypic data, CoMET tells you the
likelihoods of the data evolving through nine different evolutionary
models. For
additional information, please see the documentation.
This is the main Java class files you'll need to run CoMET.
CoMET source code (.zip) (September 2007). These are the Java source code files for CoMET.
CoMET runs
on top of the Mesquite 2.0. Please download Mesquite from http://www.mesquiteproject.org.
CoMET includes a subset of the recent version of the PAL
project.
These PAL files have been repackaged into CoMET due to the differences
between
Mesquite's PAL and CoMET's PAL files.
This guide contains a short description, written by Chunghau Lee, of CoMET plus installation and
usage instructions. Last updated 12 February 2006.
You can also see Chunghau's MS thesis. Note Chapter 3 has some simulations of punctuated models.
To cite CoMET:
C Lee, S Blay, AØ Mooers, A Singh, and TH Oakley (2006). CoMET: A Mesquite package for comparing models of continuous character evolution on phylogenies. Evolutionary Bioinformatics Online. 2:193-196. Download (pdf)
Release History
23 September 2007
Mesquite 2.0 has been released. A new version of CoMET has been released to maintain compatibility with the new Mesquite.
19 August 2006
CoMET has reverted back to using Java 1.4. Consequently, it now works under Mac OS X 10.3 and above. Many Mesquite/CoMET users still depend on OS X 10.3
The punctuated-maximal model has returned by popular demand. To use it, start with the tree window, then from the menu, select Analysis -> Tree Legend -> Tree value using character matrix (or just "Tree value using character", that works too). Then check the box that says "Show Secondary Choices" to see the option for "CoMET on Matrix (or Character): Punctuated Maximal Only". This module does not calculate the AIC, but it does print for you (in the logs) the maximal branch lengths with each of the models.
Canceling the computation of "Punctuated Avg" now works. When prompted for the asymmetry ratio, just click "Cancel".
24 June 2006
No more writing temporary files to disk when computing PuncAvg models
PuncAvg models can now deal with branch lengths that are undefined or zero
29 January 2006
Fixed bug related to adjusting the AIC's of PuncAvg models. The adjusting multiplier, dervied from the user-specified asymmetry threshold, is now properly multiplied to the ML when calculating the AIC .
Thanks to Karin Isler for reporting this bug
Fixed crashes at startup of calculations with single character and
simulator. Thanks to Karin Isler and Steve Kembel for reporting these
bugs.
19 January 2006
Updated license information and AppNote manuscript.
19 March 2005
CoMET slimmed down.
Feature added for custom punctuation definition.
Simulation output filename is now user-specified.
Updated User's Guide and AppNote
7 February 2005
CoMET for single character is included.
CoMET Simulator is included.
Website now has updated User's Guide and short paper.
7 January 2005
CoMET has been broken into proper Mesquite modules.
Results are updated as tree and characters change.
Results are written to the Mesquite log.
User's Guide is updated.
16 December 2004
Average models performance has been greatly improved! A
30-taxa
tree now takes only seconds to compute and a few kilobytes of disk
space
thanks to reduced redundancy.
17 November 2004
CoMET now prints shorter text output for results to better
fit text in the output window.
Some bad leftover code involving the Free models have been
removed.
Average models performance has been improved, but is still
slow
even for trees with 30+ taxa. The user is prompted for the option to
skip these calculations.
In order to accommodate trees larger than 255 nodes, new
tricks will
be needed for the Punctuate/Average mdoels. Current ideas include
using taxa pairing matrices instead of just pairing sequences to
reduce repeats. If the number of taxa is great enough, then these
matrices may also need to be paged to disk.