DATE: | Wednesday, October 31rst |
TIME: | 4:30 PM - 5:30 PM |
PLACE: | McConnell 320 |
TITLE: | From consensus trees to supertrees |
SPEAKER: | David Bryant, School of Computer Science, McGill University |
Phylogenetics is the study, and reconstruction, of evolutionary history. Following Darwin, this history is generally depicted by means of a leaf-labelled tree, the labels representing existant species and the internal nodes representing hypothesized ancestral species. Combinatorial problems associated with these structures, and relationships between them, have kept a community of mathematicians and computer scientists busy for some time.
I will be discussing a particular combinatorial problem: given a collection of trees how can we combine these in some systematic way into one representative tree. Constructing trees (e.g. by likelihood methods) is computationally difficult, so the facility to combine trees with overlapping leaf sets would be highly desirable, especially as part of a divide and conquer type approach. This is the supertree problem.
There is a family of methods for combining trees on the same leaf set
(consensus tree methods). I propose ways to extend these to supertree
consensus methods, and algorithmic tricks that may be employed to
construct
them.