Bridging Functional Genomics and Toxicogenomics Through DNA Microarrays In a Fish Model


Shuzhao Li

Date of Award


Degree Type


Degree Name

Doctor of Philosophy (PhD)


Coastal Sciences, Gulf Coast Research Laboratory

First Advisor

Marius Brouwer

Advisor Department

Coastal Sciences, Gulf Coast Research Laboratory


In a case study of finding gene expression signatures for environmental stressors in Cyprinodon variegates, this dissertation examines several important issues of applying DNA microarray technology to fish toxicogenomics. The most relevant disciplines, fish toxicogenomics and computational systems biology, are reviewed in Chapter 1. Chapter 2 reviews major aspects of DNA microarray technology. On DNA microarrays, even for probes that target the same transcript, large variations are seen in the probe signals. These variations are partly dependent and partly independent on probe sequences. Chapter 3 estimates the sequence independent variation by combining experimental and computational approaches. Chapter 4 and 5 take on the central problem of sequence dependent variations, that is, modeling the physiochemistry of microarray hybridization. I propose a new competitive hybridization model, which demonstrates good success on publicly available benchmark data. This new model leads the way to quantification of absolute target concentration, and brings critical insights into probe design and data interpretation of DNA microarrays. Our model relies on the accuracy of computing duplexing energy, which does yet not take into account secondary structures of probes and targets. I further explore the structural effects in Chapter 6. After one obtains microarray data, the interpretation relies on existing knowledge of functional genomics, which come mostly from model organisms other than fish. As an effort to bridge this gap, a project to construct a genome-wide fish metabolic network, MetaFishNet, is launched. MetaFishNet is based on five fish genome sequences and the latest progress in metabolic modeling, especially the two high-quality human metabolic models. Chapters 7 to 9 describe the construction process of MetaFishNet. Chapter 10 demonstrates the two roles of MetaFishNet: a tool for interpreting high throughput expression data and a systems biology framework for hypotheses generation and study design. Chapter 11 takes these methodological developments into the toxicogenomics of C. variegates. We have constructed a Cyprinodon DNA microarray, and used it to profile the gene expression of larvae exposed to hypoxia, cadmium, chromium and pyrene. The result shows that specific markers can be identified for each stressor, and stressors can be classified by transcriptomic profiles. MetaFishNet enables us to perform Gene Ontology analysis and metabolic pathway analysis on these data. "Leukotriene metabolism" and "Xenobiotics metabolism" pathways appear to be upregulated by cadmium exposure.