Rooted Triple Consensus

This page is for the implementation of the method described in Rooted Triple Consensus and Anomalous Gene Trees, Greg Ewing, Ingo Ebersberger, Heiko A. Schmidt and Arndt von Haeseler.

The program is called triplec for Triple consensus and is written in java version 1.5. So it should run without problems on Mac, Windows and *nix machines. Although the program has been tested quite thoroughly it is a work in progress and must be treated as such. In particular most of the testing and simulations were mainly done on bifurcated trees. Finally don't forget that this method works only for rooted topologies.

Licence

Triplec is free to use for commercial and non commercial purposes but comes with no warranty expressed or implied. It is provided as is because it may be useful.

Source

The source can be requested and its availability will be decided on a case by case basis. The reason is that some of the code has dependencies with a very large code base and not all of this is relevant to this program and much of this code is not yet ready for release.

Some time later this year when the rest of the code base has been properly checked for other dependencies and is ready for release, this code will be published under one of the open source licences. Most likely LGPL 2 or a BSD licence.

Download

Triplec Download

Usage

java -jar Triplec.jar NoReps InFile.newick

Here NoReps is the number of triple reconstructions to use for the consensus. This is required because reconstruction is a randomised algorithm. For small trees of about 20 taxa 1000 replicates is enough. But larger numbers should be used for bigger data sets.

InFile.newick is a text file with one newick tree per line. The trees are considered to be rooted and each tree must contain only the taxa labels of the first tree. That is trees after the first must not contain new taxa.

Although it is not a error to have trees with less taxa than the first tree as part of the consensus, this method is not a supertree method. Performance under these conditions has not been tested and we expect would perform poorly.

Tips, Memory and Performance

This code is not highly optimised as of yet. But this should not be a big constraint for most people as its performance and memory requirements are reasonably modest. Remember that by default java will only give 64 MB to a program and so on bigger trees it will seem to run out of memory. Just use the -Xmx switch in java.

Example: java -Xmx500M -jar Triplec.jar 10000 bigTrees.newick.