NCATS Home
>
News & Events
>
Features > Tox21 Challenge Winners
NCATS Announces Tox21 Data Challenge Winners
On Jan. 26, 2015, NCATS announced the winners of the Toxicology in the 21st Century (Tox21) Data Challenge 2014, a crowdsourcing competition that attracted contestants from 18 countries to design computational models to better predict chemical toxicity. Tox21 is a collaborative effort among NCATS and the National Institute of Environmental Health Sciences, Environmental Protection Agency and Food and Drug Administration to improve current methods scientists use to evaluate environmental chemicals and develop new medicines.
There are hundreds of thousands of chemicals that could be harmful to humans, and testing each one separately would be expensive and virtually impossible. Computational models — such as those submitted in response to this challenge — can predict whether a substance may be toxic by interrupting certain biological pathways based on the chemical structure. Combining these models with the knowledge already gained from Tox21 screening data can help scientists better prioritize chemicals for further toxicological testing, saving both time and money.
Challenge participants used data from nuclear receptor signaling and stress pathway assays (tests) run against Tox21’s 10,000-compound library to build models and look for structure-activity relationships. The models from the seven winning teams will become part of the Tox21 program’s arsenal of tools that help researchers assess how various chemicals might disrupt biological processes in the human body and lead to negative health effects. All displayed very good predictive power, achieving greater than 80 percent accuracy, with several models exceeding 90 percent accuracy.
“The high quality of the models gives us confidence that they can be applied to identify those chemicals that have the most potential for toxicity, enabling us to prioritize them for further evaluation,” said Anton Simeonov, Ph.D., NCATS acting scientific director. “Moreover, since the models are dependent on high-quality data and the top ones performed so well, this outcome further validates the value of our Tox21 screening efforts.”
The winners are:
| Team Name | Challenge Assay(s) | Team Member(s) | Organization(s) |
| Bioinf@JKU | Grand Challenge (all 12 assays) Stress Response Panel Aryl Hydrocarbon Receptor Nuclear Factor (Erythroid-Derived 2)-like 2/Antioxidant Responsive Element | Günter Klambauer, Ph.D.
Sepp Hochreiter, Ph.D.
Andreas Mayr, M.Sc.
Thomas Unterthiner, M.Sc. | Institute of Bioinformatics, Johannes Kepler University Linz, Austria |
| Bioinf@JKU-ensemble1 | Estrogen Receptor Alpha, Full Length Heat Shock Factor Response Element | Günter Klambauer, Ph.D.
Sepp Hochreiter, Ph.D.
Andreas Mayr, M.Sc.
Thomas Unterthiner, M.Sc.
Herbert Zaunmair | Institute of Bioinformatics, Johannes Kepler University Linz, Austria |
| Bioinf@JKU-ensemble3 | Androgen Receptor, Ligand Binding Domain | Günter Klambauer, Ph.D.
Sepp Hochreiter, Ph.D.
Ulrich Bodenhofer, Ph.D.
Andreas Mayr, M.Sc.
Thomas Unterthiner, M.Sc. | Institute of Bioinformatics, Johannes Kepler University Linz, Austria |
| Bioinf@JKU-ensemble4 | Nuclear Receptor Signaling Panel Peroxisome Proliferator-Activated Receptor Gamma | Günter Klambauer, Ph.D.
Sepp Hochreiter, Ph.D.
Birgit Hauer
Andreas Mayr, M.Sc.
Thomas Unterthiner, M.Sc. | Institute of Bioinformatics, Johannes Kepler University Linz, Austria |
| AMAZIZ | ATAD5 Mitochondrial Membrane Potential | Ahmed M. Abdelaziz Sayed | Technical University of Munich |
| Dmlab | Androgen Receptor, Full Length Aromatase p53 | Gergő Barta, M.Sc. | Budapest University of Technology and Economics |
| Microsomes | Estrogen Receptor Alpha, Ligand Binding Domain | Yoshihiro Uesawa, Ph.D. | Department of Clinical Pharmaceutics, Meiji Pharmaceutical University |
Learn more about the Tox21 Data Challenge.
Posted January 2015
Social Media Links