Chemical Hazard Estimation and Method Comparison with OWL-Encoded Toxicity Decision Trees

Industrial and regulatory evaluation of chemical toxicity is often done via statistical analysis of chemical features focusing on chemical structure and function. One popular method to characterize chemical toxicity involves the development of decision trees based on large sets of empirical toxicological data where chemicals are assigned toxicity or activity classes. In this paper, we describe the representation of decision trees as OWL ontologies that can be used to carry out initial evaluation of toxicity and activity of prospective chemical products. We further discuss how trees derived from different datasets can be semantically compared by examining the logical equivalence of the toxicity and bioactivity classes in different trees. Taken together, this initial work forms the basis for continued investigation into OWL-driven semantic framework for toxicity evaluation.