Modeling gene regulatory networks

COMP 232

Peng Zhang, zhangpe@umdnj.edu, Ming Ouyang, and William J. Welsh. Department of Pharmacology, University of Medicine & Dentistry of New Jersey, Robert Wood Johnson Medical School, 675 Hoes Lane, Piscataway, NJ 08854
We present a differential equation modeling method that infers the gene regulatory network from time course expression data in human prostate cancer cells. We propose several methods that focus on co-regulated gene groups. A stability-based clustering algorithm is used to detect and partition all the differentially expressed genes into an optimum number of clusters so that each cluster represents a unique structure feature of the data that corresponds to a co-regulated gene group. The centroid value of each cluster is used to represent the expression level of that cluster at each time point. Multiple regression is used to fit the cluster centroids with a set of differential equations. The change in expression level of each cluster centroid is a weighted sum of expression levels of all cluster centroids. The resulting interaction matrix is statistically tested by permutation analysis to provide a final image of regulatory networks among the clusters.
 

Poster Session -- Sponsored by Novartis Institutes for BioMedical Research
6:00 PM-8:00 PM, Tuesday, 30 August 2005 Washington DC Convention Center -- Hall A, Poster

Sci-Mix
8:00 PM-10:00 PM, Monday, 29 August 2005 Washington DC Convention Center -- Hall A, Sci-Mix

Division of Computers in Chemistry

The 230th ACS National Meeting, in Washington, DC, Aug 28-Sept 1, 2005