Defense of Thesis Proposal - Chengwei Su

Dartmouth Events

Defense of Thesis Proposal - Chengwei Su

A Novel Algorithm for Efficient Learning of Bayesian Networks of Disease Susceptibility Using Prior Knowledge

Wednesday, October 16, 2013
10:00am-11:00am
Jackson Conference Room, Cummings Hall
Intended Audience(s): Public
Categories:


“A Novel Algorithm for Efficient Learning of Bayesian Networks of Disease 
Susceptibility Using Prior Knowledge”
 

 
Thesis Committee
Mark Borsuk, Ph.D. (Chair)
George Cybenko, Ph.D.
Casey Greene, Ph.D.
Yulei He, Ph.D.


 
Abstract
 
 
Uncovering the genetic basis for cancer and deciphering the relative contribution of environmental exposure are critical steps toward the goal of developing a personalized understanding of environmental health risk. The primary objective of the proposed research plan is to develop and validate the approach, algorithms, and software for efficiently learning this complex web of relations from a combination of biological knowledge, published literature, and genetic epidemiological data. Automated learning of Bayesian network structure has been shown to be a promising approach to this task. However, application of existing structural learning algorithms to genome-wide data is limited by the large number of possible interactions among candidate genetic polymorphisms. Thus, a specific aim of the proposed project is to develop and test intelligent strategies for prioritizing the structural search space using prior information.  The process of retrieving the relevant information from publicly-available databases and literature archives will be automated through enhancement and utilization of the existing Integrative Multi-species Prediction (IMP) tool. The resulting integrated bioinformatics approach will be exemplified through application to an extensive genetic epidemiological data set from a large, population based, case-control study of bladder cancer in New Hampshire. By combining recent advances in the development of Bayesian networks, innovative methods for capturing prior knowledge, and rigorous simulation-based validation, it is anticipated that the proposed research will yield new avenues for deciphering how genes and the environment interact to determine cancer risk. Identifying these key combinations of risk factors is critical to the successful implementation of targeted screening programs and optimal use of interventions such as chemopreventive agents or exposure cessation to avert disease among susceptible individuals.

For more information, contact:
Daryl Laware

Events are free and open to the public unless otherwise noted.