Bryan Kolaczkowski
bkolaczk@cs.uoregon.edu

University of Oregon
Institute of Neuroscience
1254 University of Oregon
Eugene, OR 97403-1254 USA

Advisors:
Dr. John Conery, Computer & Information Science
Dr. John H. Postlethwait, Institute of Neuroscience
Dr. Joe Thornton, Ecology & Evolutionary Biology

________________________

Research Interests

My research interests are centered in the study of Bioinformatics as a tool for addressing both biological and computational questions. Through the merging of computational science and biological information studies, complex problems in ecology, evolution, and genomics can be addressed while developing novel approaches to automating efficient large-scale scientific inquiry. This approach can benefit both biological and computational science by uncovering new problems and generating new information which leads to a larger understanding of the complexities of our natural world and our approaches to understanding it.

Current research has focused on performance analysis of parametric and nonparametric phylogenetic reconstruction methodologies under heterogeneous evolutionary models. Parametric phylogenetics (i.e. maximum likelihood and Bayesian analysis) has incorporated a great deal of knowledge regarding the molecular evolutionary mechanisms at play in the real world. However, these methods -- as currently implemented -- assume the molecular evolutionary model to be homogeneous both across lineages and (to a large extent) across characters in an alignment matrix. There is ample evidence to suggest that this assumption may be false. The purpose of this research is to explore how violations of model homogeneity may impact performance of parametric and nonparametric (i.e. maximum parsimony) phylogenetic reconstruction.

Computer simulation studies have a history of being used to address phylogenetic method performance, and we have extended these studies by incorporating model heterogeneity into the evolutionary models used to derive simulated sequence alignments. To facilitate the creation, curation, and analysis of molecular sequences generated under complex heterogeneous models, we have constructed a novel system centered around a MySQL relational database. Using custom software written in C++, we are able to automatically generate thousands of sequence alignments across a range of interesting parameter values, pipe these alignments to existing phylogenetic analysis software (including PAUP* and MrBayes), collect resulting reconstructed tree topologies, and score results. This system has allows large-scale simulation studies to be conducted with minimal human intervention, allowing us to ask both larger and more complex questions. It is believed that this work may not only indicate areas of parameter space where model heterogeneity matters but also lead to interesting novel phylogenetic method development in the future. The software system developed for this research may also serve as a model for future automation of bioinformatic studies in comparative biology.

________________________

Publications

... none currently ...