« Back to abstracts page

EPA'S Virtual Embryo: Modeling Developmental Toxicity

Thomas Knudsen National Center for Computational Toxicology US EPA, Research Triangle Park, NC

Authors

Thomas Knudsen

Abstract

The EPA ToxCast™ research program uses a high-throughput screening (HTS) approach for predicting the toxicity of large numbers of chemicals. Phase-I tested 309 well-characterized chemicals (mostly pesticides) in over 600 assays of different molecular targets, cellular responses and cell-states. The goal of the present study was to build predictive signatures for prenatal developmental toxicity from altered molecular targets. Developmental defects are the result of complex interplay between environmental and genetic factors and likely involves the direct or indirect disruption of conserved morphogenetic pathways patterning the embryo. To help navigate this complexity, EPAs Virtual Embryo project is building agent-based models that simulate pathway-level perturbations and network state relations for critical developmental processes and toxicities. Predictive signatures of developmental toxicity were identified in EPA's ToxMiner™ database. Machine learning techniques returned 232 significant multivariate associations between molecular targets and fetal endpoints. These included 70 non-redundant molecular targets and 12 cellular effects mapping to 24 Ingenuity pathways, 28 KEGG pathways, and 48 OMIM disease phenotypes. Further analysis linked developmental toxicity with 35 molecular targets in the rabbit and 48 in the rat; the strongest linkages were to defects of the eye, kidney ureter, and palate. Computational (in silico) systems models are being built in CompuCell3D and Python to simulate the effects for various perturbations in a virtual embryo for predictive modeling. [This work does not reflect Agency policy].