Session 3: High-Throughput Testing Data

The US EPA's ToxCast Project: A Public Resource for Environmental Chemical High-Throughput Testing Data

Keith A. Houck, Ph.D.


National Center for Computational Toxicology, US EPA


Cell and Molecular Biologist


Safety assessment of drugs and environmental chemicals relies extensively on animal testing. However, the quantity of chemicals needing assessment and challenges of species extrapolation drive the development of alternative approaches.

The EPA’s ToxCast and the multiagency Tox21 programs address this through use of an extensive in vitro screening program to generate data on a large library of important environmental chemicals. These in vitro assays encompass biochemical formats targeting proteins that may be potential molecular initiating events, cellular assays that provide coverage of critical signaling pathways and toxicity phenotypes, and model organism assays for developmental processes.

The resulting heterogeneous data sets are entered in to a robust data analysis pipeline that uses standardized data formats to process, normalize, model, qualify, flag, inspect, and visualize the data.

A major achievement of the program is making these data publically available. Full data matrices including raw data and R code for running the pipeline are available for download from EPA’s National Center for Computational Toxicology’s website.

We have also developed a data browsing and visualization tool called the iCSS Dashboard to access data in a selective manner. Data generated from the Tox21 program are deposited in PubChem in a timely fashion. It is hoped that ready public access to these testing data will contribute to development of more robust predictive toxicology models.

This work does not necessarily reflect official Agency policy.