Reverse Engineering of Gene Regulatory Networks for Discovery of Novel Interactions in Pathways Using Gene Expression Data


Tanwir Habib

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Biological Sciences

First Advisor

Mohamed O. Elasri

Advisor Department

Biological Sciences


A variety of chemicals in the environment have the potential to adversely affect the biological systems. We examined the responses of Rat ( Rattus norvegicus ) to the RDX exposure and female fathead minnows (FHM, Pimephales promelas ) to a model aromatase inhibitor, fadrozole, using a transcriptional network inference approach. Rats were exposed to RDX and fish were exposed to 0 or 30mg/L fadrozole for 8 days. We analyzed gene expression changes using 8000 probes microarrays for rat experiment and 15,000 probe microarrays for fish. We used these changes to infer a transcriptional network. The central nervous system is remarkably plastic in its ability to recover from trauma. We examined recovery from chemicals in rats and fish through changes in transcriptional networks. Transcriptional networks from time series experiments provide a good basis for organizing and studying the dynamic behavior of biological processes. The goal of this work was to identify networks affected by chemical exposure and track changes in these networks as animals recover. The top 1254 significantly changed genes based upon 1.5-fold change and P < 0.05 across all the time points from the fish data and 937 significantly changed genes from rat data were chosen for network modeling using either a Mutual Information network (MIN) or a Graphical Gaussian Model (GGM) or a Dynamic Bayesian Network (DBN) approach. The top interacting genes were queried to find sub-networks, possible biological networks, biochemical pathways, and network topologies impacted after exposure to fadrozole. The methods were able to reconstruct transcriptional networks with few hub structures, some of which were found to be involved in major biological process and molecular function. The resulting network from rat experiment exhibited a clear hub (central in terms of connections and direction) connectivity structure. Genes such as Ania-7, Hnrpdl, Alad, Gapdh, etc. (all CNS related), GAT-2, Gabra6, Gabbr1, Gabbr2 (GABA, neurotransmitter transporters and receptors), SLC2A1 (glucose transporter), NCX3 (Na-Ca exchanger), Gnal (Olfactory related), skn-la were showed up in our network as the 'hub' genes while some of the known transcription factors Msx3, Cacng1, Brs3, NGF1, etc. were also matched with our network model. Aromatase in the fish experiment was a highly connected gene in a sub-network along with other genes involved in steroidogenesis. Many of the sub-networks were involved in fatty acid metabolism, gamma-hexachlorocyclohexane degradation, and phospholipase activating pathways. Aromatase was a highly connected gene in a sub-network along with the genes LDLR, StAR, KRT18, HER1, CEBPB, ESR2A, and ACVRL1. A credible transcriptional network was recovered from both the time series data and the static data. The network included transcription factors and genes with roles in brain function, neurotransmission and sex hormone synthesis. Examination of the dynamic changes in expression within this network over time provided insight into recovery from traumas and chemical exposures.