Inference of Gene Regulatory Network Based on Gene Expression Dynamics in Response to Environmental Signals

Open Access
Wang, Yaqun
Graduate Program:
Doctor of Philosophy
Document Type:
Date of Defense:
June 05, 2015
Committee Members:
  • Rongling Wu, Dissertation Advisor
  • Rongling Wu, Committee Chair
  • Runze Li, Dissertation Advisor
  • Runze Li, Committee Chair
  • Zhibiao Zhao, Committee Member
  • Donna Coffman, Committee Member
  • Gene Regulatory Network
  • Gene Expression
  • Gene Cluster
  • DBN
  • Ordinary Differential Equation
  • Gene-Environment Interaction
Thousands of genes are encoded on the genome and their products play important roles to cell survival, phenotypic characteristics of organisms and adaptive behaviors of organisms when environment changes. Detecting of particular sets of genes whose expressions are adaptive in response to environmental signals and identification of dynamic gene regulatory networks (GRN) can help us to understand the mechanistic base of gene-environment interactions and gene-gene interactions in a systematic way. However, it is a challenging work to analyze gene expression across two-dimensional spaces, time and environmental state. In this dissertation, we develop a functional clustering framework based on a mixture model to analyze time-course gene expression. The mathematical aspects of gene expression dynamics have been captured by Legendre polynomial and the impact of environment on gene expression has been considered jointly. We outline a number of quantitative testable hypotheses about the patterns of dynamic gene expression in changing environments and gene-environment interactions causing developmental differentiation. The method is illustrated with simulation studies and application on a real data set from a rabbit hemodynamic study. In addition, we propose two models for inference of GRN based on gene expression. We reform the Dynamic Bayesian Network (DBN) model for identification of GRN to overcome its limitation that evenly spaced measurements is required. The reformed model can accommodate to any possible irregularity and sparsity of time-course expression data by adaptively fitting gene expression curves, followed by a step of interpolating data at missing time points before conducting of DBN analysis. We also develop an ordinary differential equation (ODE) model to reconstruct GRNs based on functional clustering of genes. A set of ordinary differential equations are constructed to quantify the dynamic of GRN and the regulatory effects including positive and negative regulation are identified in a regression setting by using Smoothly Clipped Absolute Deviation (SCAD)-based variable selection. Both GRN models are equipped with unique power to integrate gene expression data from multiple environments and, therefore, provides an unprecedented tool to elucidate a comprehensive picture of GRN. By analyzing real data sets from a surgical study and through extensive simulation studies, the new models have been well demonstrated for their usefulness and utility.