| Kim Lab of Computational Evolutionary Biology | |
| Public Private Project1 Project2 Project3 Project4 Archive | ||
|
Home People Projects Publications Downloads Cluster Jobs Discussions Biology Department School of Arts and Sciences University of Pennsylvania
103I Lynch Laboratory 433 S University Avenue Philadelphia, PA 19104 USA off: (215) 746-5187 lab: (215) 898-8395 fax: (215) 898-8780 email: junhyong@sas.upenn.edu |
KevinBullaugheyProjects: I'm working on two projects: one for my biology thesis, and the other for my CSE thesis. Both involve phylogenetic trees. CSE Thesis: Here's a copy of my abstract in progress: Many problems in computational biology make use of phylogenetic trees to discuss the likely ancestral relationships among DNA sequences. In order to make useful statistical inferences from phylogenetic tree comparison metrics, one must have well characterized distributions of the comparison metrics relative to various null models for random tree sampling. For statistics that compare tree similarity, such as Maximum Agreement Sub-Tree (MAST), only the asymptotic behavior of the distribution is well characterized. However, the 'tails', or the regions of the distribution of least frequency, have not yet been well characterized. It is the purpose of this project to investigate the statistical nature of the tails of the distribution of MAST scores on random trees. Because the number of labeled trees increases combinatorially with the number of taxa, it is impossible to generate all possible trees for large numbers of taxa. Therefore simulations of various tree-generating processes or conditional probabilities may be used to enhance sampling bias. Biology Thesis:
I intend to do a simulation study of two phylogenetic tree reconstruction methods under two different models of evolution. The two tree reconstruction methods are maximum likelihood and maximum parsimony. The two models of evolution are the i.i.d. site model and a model where the sequence space is limited by the constraints of context-free-grammars. The idea behind the second model is to select a random tree and evolve a grammar over that tree so at the end of the evolutionary process we have one grammar at each leaf. Each grammar then emits one or more sequences consistent with that grammar. The question is how do the tree reconstruction methods perform on the limited sequence space compared to the i.i.d. model. Contact Information Address: 2300 Pine Street Apt 1 Philadelphia, PA 19103 Mobile: 610-656-3755 AIM: kbullaughey email: bullaugh@sas.upenn.edu Majors: Biology (computational concentration) Computer Science Engineering Originally from West Chester, PA. I'll be graduating May 2004, so if you need me after that, try getting updated contact info from my parents at 610-793-2370.
| |