|
|||||||||||||||
 |
Graduate Seminar2005-2006 Schedule2004-2005 Schedule - Spring 04 Schedule - Fall 2003 Schedule
AbstractsTom Smith - Gibbs Sampling & Phylogeny in Gene Regulation Analysis Abstract: Gene expression is the process of creating a protein from the DNA sequence of a gene. Various regulatory mechanisms control gene expression and determine which proteins are produced in which cells. Signals for these mechanisms are often encoded in DNA sequences known as regulatory elements. The discovery of such regulatory elements greatly contributes to our understanding of cell function. The recent availability of complete genome DNA sequences from many species offers new opportunities for computational analysis. I am currently developing a new method to discover regulatory elements in multiple genomes simultaneously. My algorithm leverages information about evolutionary relationships between species to predict the locations of regulatory elements. I will present this new algorithm and share preliminary results of its application to bacterial genomes. I will also briefly discuss a software framework which I developed to implement this algorithm generically. No knowledge of molecular biology or bioinformatics will be assumed. Alessandro Assis - Adapting Educational Hypermedia according to the Learner Interaction Behavior Abstract: Adaptive Hypermedia is a field that studies how to individualize interaction with hypermedia systems according to user characteristics such as preferences, interests and needs. Education is one of the areas where it has been mostly applied. Researchers discuss how to adapt web-based learning systems according to individual learning traits. We present a system named ADaPtor that adapts educational materials for a particular goal. ADaPtor aims to optimize learning effectiveness and efficiency by iteratively adapting a learning cycle according to the learner interaction behavior. A learning cycle is grounded on instructional design and learning style theories. When interacting with a learning cycle, a learner finds instructional content attractive to his/her particular learning style but also interacts with instruction formatted to satisfy different learning styles. ADaPtor's goal is optimizing learning performance by individualizing the learning cycle experience. Experimental results will be presented and discussed. Abstract: Many recent biological techniques, such as PCR primer design and molecular beacon design, rely on the ability to predict hybridization thermodynamics for nucleic acids accurately. A typical problem involves two RNA or DNA sequences in solution, which might hybridize to form heterodimers or homodimers, or fold as single strands. In this talk, I will describe the design and implementation of a computational model to predict the concentrations of each of these species, as well as quantities such as enthalpy/entropy and UV absorption. Jeffrey Baumes - Algorithms for Discovering Hidden Groups in Communications Abstract: The goal of this research is to develop statistical and algorithmic methods and software tools which discover groups which do not advertise their existence in communications data. The algorithms and tools will aid authorities in countering future terrorist threats by detecting hidden groups in their planning stages. These approaches and systems will also assist sociologists in understanding social group structure. I will present algorithms which discover groups which exhibit different types of correlation over time. I will also discuss a new definition of clustering and novel overlapping clustering algorithms which are applicable to a number of areas in computer science such as data mining and bioinformatics, where in the past only partitioning tools have been available. A software system SIGHTS incorporates these algorithms into a single GUI framework. |
|
|||||||||||||
News
