Towards in silico toxicity prediction: Analyzing superfund chemicals for their biological properties using the Toxcast data
Nguyen, Andrew · Carolina Digital Repository (University of North Carolina at Chapel Hill) · 2019
In 1980, the Superfund program was established in response to the growing concern about hazardous waste sites in the United States. Many toxicological exposure assessments, remediation processes and estimation of human health risks at Superfund sites depend on animal studies as a model for assessment. However, with the existence of thousands of potentially harmful toxicants, using a traditional in vivo approach to prioritize chemicals can become time-consuming and expensive. In this project, we set out to prioritize chemicals found at U.S. Superfund sites by incorporating a novel computational toxicological modeling tool, ToxPi, and half-maximal activity (AC50) data from in vitro assays run by the ToxCast program. Focusing on the biological processes (n=11) defined by ToxCast, we defined overall biological potency profiles, derived a rank based on a score for the 244 SPL toxicants which had a statistically significant correlation to the ATSDR 2013 SPL Rankings, and identified unique bioactivity trends.