Algorithms and Software for Phylogenetic Supertrees



Sanderson and Gusfield

Supertrees are phylogenies (rooted evolutionary trees) assembled from smaller phylogenies that share some but not all taxa (twigs or end branches) in common. Thus, supertrees can make novel statements about relationships of taxa that do not co-occur on any single source tree while still retaining all the hierarchical information from all source trees. As a method of combining existing phylogenetic information, supertrees potentially solve many of the problems incurred by other methods, such as lack of homologous characters, incompatible data types, or combining trees with different taxa. However, despite the potential for constructing large trees and synthesizing phylogenetic relationships from trees based on disparate sources of evidence, supertree construction has mainly been an informal "by hand" operation, done without the use of specially developed software. Furthermore, there is little ground for assessing the reliability of such supertrees.

In a collaboration among biologists and computer scientists, Drs. Sanderson, Gusfield, and colleagues are examining a new class of methods that is somewhat analogous to phylogenetic distance methods but uses a novel distance measure based on "flips" of cells in the matrix representation of the source trees. In addition to generating supertrees, flip-supertree methods will help systematists focus on poorly known or contentious taxa, thereby directing future systematic effort where it is most needed. Accuracy and speed of supertree methods will be tested through simulation studies, and through use of the web-based database of phylogenetic trees in TreeBASE, to determine methods most useful to systematists. The graphically oriented software to be developed will implement various supertree algorithms and offer improved methods to visualize and compare trees that share some, but not all, taxa in common.

2000-10-01 to 2004-09-30
Award Amount: 
Award Number: