Humans are exposed to various chemical agents through food, cosmetics, pharmaceuticals and other sources. Exposure to chemicals is suspected of playing a main role in the development of some adverse health effects in humans. Additionally, European regulatory authorities have recognized the risk associated with combined exposure to multiple chemicals. Testing all possible combinations of the tens of thousands environmental chemicals is impractical. This PhD project was launched to apply existing computational systems biology methods to toxicological research.
In this thesis, I present in three projects three different approaches to using computational toxicology to aid classical toxicological investigations. In project I, we predicted human health effects of five pesticides using publicly available data. We obtained a grouping of the chemical according to their potential human health effects that were in concordance with their effects in experimental animals. In project II, I profiled the effects on rat liver gene expression levels following exposure to a 14-chemical mixture ± the presence of an endocrine disrupting chemical. This project helped us shed light on the mechanism of action of the 14-chemical mixture and the endocrine disrupting chemical. In project III, I modeled a predictive signature for an in vivo endpoint that is sensitive to endocrine disruption. I used publicly available data generated for the purpose of modeling predictive signatures for various in vivo endpoints. From this modeling effort, I have suggested a mechanism of action for a subset of the chemicals that has not previously been associated with endocrine disruption.
The use of computational methods in toxicology can aid the classical toxicological tests by suggesting interactions between separate components of a system thereby suggesting new ways of thinking specific toxicological endpoints. Furthermore, computational methods can serve as valuable input for the hypothesis generating phase of the preparations of a research project.