Available software for analyzing phylogenetic trees, written by Stephen J. Willson:
Software is available to build phylogenetic trees from data sets. For each collection of four taxa, the best tree for those species alone is computed either by the criterion of Maximum Parsimony or by the criterion of Higher Order Parsimony, which introduces a correction to Maximum Parsimony that reduces the effect of long-branch attraction. If there is a large gap between the treelength of the best tree and the runner-up tree, we have high confidence in the correctness of that tree. Data about all the trees for four species at a time are placed in a list. We seek to find a global tree by combining together the best trees for each group of four taxa, using primarily the trees in which we have high confidence.
Optionally, the user may also utilize an error-correction process which modifies the trees previously selected for each quartet so as to increase the degree of consistency among the quartets. This process usually strengthens the signal for the correct tree.
There are two groups of software.
Building rooted supertrees using distances
Suppose that a collection of trees Ti with overlapping leaf sets Xi is proposed. A supertree will be a single tree T with leaf set the union of the Xi which tells the information in Ti whenever T is restricted to Xi. The program BUILDWITHDISTANCES produces proposed supertrees in a fast manner. The input trees are required to have distance information. For certain pairs (a,b) of taxa, input is made of l(a,b), which is the length in a tree Ti of the path from a to lca(a,b), the lowest common ancestor of a and b.
The following files are available:
Focusing on use of the highest signal strength.
For each collection of four taxa, the best tree for those species alone is computed either by the criterion of Maximum Parsimony or by the criterion of Higher Order Parsimony, which introduces a correction to Maximum Parsimony that reduces the effect of long-branch attraction. The software then builds the phylogenetic tree on the basis of local inconsistency measures. For each placement, it measures the local inconsistency, and it chooses the placement with the strongest signal for correctness. It also permits interactive building of the phylogenetic tree. The topology for each quartet is selected using either maximum parsimony or higher order parsimony, a corrected version of maximum parsimony that reduces long-branch attractions, at the option of the user. The software is written in C or C++. It has worked on a Power Macintosh 7300/200 as compiled by Codewarrior 9.
The following files contain the source code in C, sample data sets, and instructions:
Executable files for the Power Macintosh are also available:
For more information about my work on building phylogenetic trees, click
For more information about the Laurence H. Baker Center for Bioinformatics and Biological Statistics click CBBS
For my homepage, click
Send comments to me by e-mail at firstname.lastname@example.org
Last revised March 29, 2004