Fordham University
Department of Computer & Information Science
and
Interdisciplinary Faculty Seminar on Biomedical Informatics
Present:

Date : Thursday, February 22, 2007
Time : 6:00pm-7:00pm
Location : Leo Lowenstein Building, LL 419

Abstract:
The reconstruction of the tree of evolutionary history, or phylogeny, of species usually involve computationally intensive search algorithms in the space of all phylogenies. Such analyses often produce thousands of candidate trees; biologists resolve the conflict by computing the consensus of these trees such as the strict and majority consensus methods. These methods all produce only one consensus tree as a summary of common features and omit conflicting ones. Though efficient as a means to reduce information, this inherent limitation means single-tree consensus methods can be unsatisfactory when the tree space "landscape" is complex. To address this issue, we proposed an alternative approach by using clustering algorithms on the set of candidate trees.

In the first half of my talk, I will briefly review the problem of phylogeny reconstruction and common algorithms for tree space search and consensus. I will then cover the frame work of clustering in the tree space as a generalized consensus method. We propose bicriterion problems, in particular using the concept of information loss, and new consensus trees called representative trees that minimize the information loss. Our empirical study using four biological datasets shows that our approach provides a significant improvement in the information content with only a small amount of overhead. Furthermore, the consensus trees we obtain for each of our large clusters are more resolved than the single-tree consensus trees.

Bio:

Dr. Li-San Wang received an M.S. and a Ph.D. degree in Computer Science from the University of Texas at Austin in 2000 and 2003. He was postdoctoral fellow at the University of Pennsylvania from 2003 to 2006. Since 2006, he is an Assistant Professor of Pathology & Laboratory Medicine in the School of Medicine, University of Pennsylvania. His research areas include disease-related bioinformatic research, computational methods for microarray analysis and systems biology, and computational phylogenetics. Dr. Wang has served as program committee member of EITC/Bioinformatics 2004, RECOMB Satellite Workshop on Comparative Genomics and as referee for Journal of Computational Biology, Mathematical Biology, Systems Biology and RECOMB.

For more information or directions, contact Ms. Diane Roche (718) 817-4480