Computer science announces student poster contest winners

The Department of Computer Science and Engineering held a student poster contest during its Fall 2010 Industrial Affiliates Program Meeting Sept. 14.Suzanne MatthewsMore than 20 posters were submitted for this event. Suzanne Matthews won first place in the contest for her poster, "An Efficient and Extensible Approach for Compressing Phylogenetic Trees." Matthews' adviser is Dr. Tiffani Williams.Matthews is a Ph.D. candidate in the department whose research focuses on creating novel algorithms to analyze and manage large sets of phylogenetic trees."A "phylogenetic tree is a structure that depicts how a group of organisms (taxa) are related," Matthews said. "Phylogenetic trees have several applications, from ecological preservation to forensics and drug discovery. In the process of inferring a phylogeny, popular methods tend to produce collections of tens to hundreds of thousands of equally plausible trees. As these tree collections grow large, they become very costly to store. While standard compression methods can compress trees, they operate by exploiting redundancy at the string level. If there is no apparent redundancy of the strings in a collection of trees, standard compression methods become ineffective."TreeZip looks for redundancy at the topological (or tree) level of a set of phylogenetic trees. This allows it to uniquely identify shared evolutionary relationships between a set of trees. Unlike 7zip, TreeZip is robust to branch rotations, allowing it to consistently achieve excellent compression. 7zip's efficacy decreases with the level of branch rotations in the file. TreeZip also provides the capability of semantic decompression. In addition to decompressing the original file, Treezip can also decompress trees of semantic interest to the user, such as the set of unique trees or the consensus tree. This extensibility highlights TreeZip's power, and is another feature not reproducible by standard compression methods By itself, TreeZip can compress a file containing a collection of trees by at least 72 percent of its original size. When combined with 7zip, it is the most effective method for compressing phylogenetic trees, compressing tree files on average by 92 percent."We believe that TreeZip will be a vital compression method in the biological community, allowing for the easier exchange and storage of tree data with colleagues around the world. We also believe the TreeZip algorithm has implications for compressing other types of tree-like data, such as XML files and graphs."Matthews says her ultimate goal is to design new methods that will be useful in the construction of the Tree of Life, a phylogenetic tree that will depict the relationships among the estimated 10 to 100 million species on the planet."Reconstructing the Tree of Life," said adviser Williams, "is the computational grand-challenge of evolutionary biology."Also placing in the contest was Grant Bammer who won second place for his poster "Using Decision Tree Learning to Study the Convergence of Phylogenetic Analysis." Bammer's adviser is also Williams. The third-place award went to Drew Logsdon for his poster, "Hand Gesture Tool Manipulation in a Gaming Environment." Logsdon is advised by Dr. Tracy Hammond.