Genome-Wide In-Silico Modeling of Liver Gene Regulatory NetworksEPA Grant Number: FP91650
Title: Genome-Wide In-Silico Modeling of Liver Gene Regulatory Networks
Investigators: Gatti, Daniel M.
Institution: University of North Carolina
EPA Project Officer: Jones, Brandon
Project Period: September 1, 2008 through August 31, 2011
RFA: STAR Graduate Fellowships (2008) RFA Text | Recipients Lists
Research Category: Academic Fellowships
An understanding of how genetic polymorphisms, gene expression and toxicity are related in the liver will increase our ability to predict both which environmental chemicals will prove harmful and which human sub-populations are particularly vulnerable to such chemicals. Building gene regulatory networks is one of the required linkages in constructing the source-to-outcome continuum. This clearer understanding of how gene expression varies and is regulated in a genetically diverse population will aid in predicting which sub-populations may have an increased risk of injury due to exposures that might otherwise prove benign. We propose to use measurements of constitutive gene expression in several panels of inbred mice combined with statistical analyses of transcription factor activity to construct robust networks that regulate gene expression in the liver. Further, the techniques developed in this project can be applied to human data sets.
We propose to use mouse models for the development of computational techniques to find robust transcriptional networks which vary in a genetically heterogeneous populations and to produce candidate gene expression networks for future use in predictive models of liver toxicity. Panels of inbred mouse populations have been shown to model the range of phenotypic variation observed in human populations. Using existing data from two separate panels of inbred mice, we will perform quantitative trait locus (QTL) mapping on the gene expression measurements. By extracting the reproducible QTLs across the two data sets, we will find networks of correlated genes which are likely to share common regulatory control. Lastly, we will combine this data with transcription factor activity from a variety of sources and use statistical models to construct liver gene expression networks.
The first part of this project will produce fast QTL mapping software in order to handle the volume of QTL mapping that will be performed. The prototype of this tool is an order of magnitude faster than existing tools and will be publicly available to the community as a friendly, graphically driven software package. The second part of this project will mine two existing murine liver gene expression data sets to find networks of highly correlated genes which share common regulation loci in the genome. Lastly, we will combine these gene expression networks with transcription factor activity into a statistical model which will produce the most likely candidate networks for inclusion in future predictive models of liver toxicity and serve as primary candidate networks for in vitro follow-up experiments.