Research Abstracts Online
January - December 2011
University of Minnesota Twin Cities
School of Public Health
Division of Biostatistics
PI: Yen-Yi Ho
nPARS: A Comprehensive Search Algorithm for Constructing Bayesian Networks Using Large-Scale Genomic Data
Gene association studies have reported on genomic locations that could affect susceptibility to human disease. However, many mechanisms remain poorly described. This project considers analytic methods for gene association study design, which integrate information about genomic DNA variations with gene expression, to discover genetic networks that are associated with phenotypic outcomes. The researcher is developing a comprehensive search approach named Network Partition Reassembly Search (nPARS), to construct Bayesian networks using large-scale genomic data. Unlike other commonly used search algorithms, nPARS is designed to consider all the variables in a given large-scale genomic data set during the search process. The next step is to demonstrate its performance and usage by investigating the proposed procedure in ve-node networks. This will involve a simulation study to compare nPARS with two other search algorithms, exhaustive search (Exh) and a greedy search with random restarts (Greedy), assuming various underlying five-node network scenarios.