NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge Opens Today

Finding better ways to predict the toxicity of chemicals

DREAM Conference 2013

SEATTLE--()--An innovative crowdsourced computational Challenge, called the NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge, launches today. The objective of this Challenge is to obtain a greater understanding about how a person’s individual genetics can influence cytotoxic response to exposure to widely used chemicals. It is being led and organized by scientists from Sage Bionetworks, DREAM, the University of North Carolina, the National Institute of Environmental Health Sciences (NIEHS) and the National Center for Advancing Translational Sciences (NCATS).

Challenges such as the NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge engage diverse communities of scientists to competitively solve a specific problem in a given time period by placing scientific data, tools, and the resulting predictive models into an open Commons or workspace – in effect, “crowdsourcing” data analysis.

Those interested to participate in this Challenge can sign up here: https://www.synapse.org/ - !Challenges:DREAM8. The Challenge will close on September 15, 2013, and the top-scoring team(s) will be announced at the November, 2013 DREAM Conference (www.iscb.org/recomb-regsysgen2013) taking place in Toronto, Canada.

The NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge

The NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge represents the type of Challenge that Sage Bionetworks and DREAM are most interested to run: namely those with the potential to provide powerful scientific insights and meaningful public impact. Toxicity testing that monitors health risks posed to humans through chemical exposure is a crucial component of public health. Yet currently, for every chemical that has been tested for toxicity, there are thousands that remain as yet untested. To address this, toxicologists are highly interested to leverage the dramatic technological advances in molecular biology and computer science that now make it possible to use high throughput in vitro biochemical- and cell-based assays and genomic data for toxicological testing. Towards this goal, the NIEHS/NCATS/UNC team recently conducted the largest ever population-based in vitro cytotoxicity study by treating 1086 human lymphoblastoid cell lines representing 9 distinct geographic subpopulations (made available via the 1000 Genomes Project: www.1000genomes.org), with 179 pharmaceutical and environmental chemicals. The resulting cytotoxicity data when paired with the publicly available genetic and genomic data on each of the respective cell lines provides a unique dataset that researchers can use to predict toxic responses to chemical compounds across a genetically diverse human population.

“Predicting how different people or groups of people will respond to certain chemicals is difficult to determine, but important for protecting the public’s health,” said Raymond Tice, Ph.D., who heads the Biomolecular Screening Program at NIEHS. By positioning this data for a DREAM Challenge, a community of Challenge participants will be asked to solve one or both of two related sub-Challenges: (1) Use the data to develop a model that accurately predicts individual responses to compound exposure based on genomic information and (2) Use the data to develop a model that accurately predicts how a particular population will respond to certain types of chemicals.

“We are delighted to partner with Sage/DREAM to release this unique dataset obtained through a broad partnership with NIEHS and NCATS,” said Ivan Rusyn, M.D., Ph.D., professor of environmental sciences and engineering at UNC’s Gillings School of Global Public Health. “The long-term strategic value of accurate predictive models will be invaluable for both protection of human health and the environment, and support of innovations in the chemical industry.”

“The collaboration with Sage/DREAM is an important extension of our ongoing partnership with NIEHS and UNC,” added Anton Simeonov, Ph.D., NCATS acting scientific director of discovery innovation. “We have capitalized on NIEHS’ expertise in toxicology, UNC’s expertise in genomics and NCATS’ quantitative high throughput screening technology platform to evaluate thousands of chemicals at multiple concentrations.”

Three-month Challenge period with continuous participation

Sage and DREAM’s organizers plan to deploy tools and incentives throughout the three-month Challenge period to stimulate a high level of continuous participation. For example, within a month of opening this Challenge, organizers will go live with a real-time leaderboard for one of the sub-Challenges: this leaderboard will post the “scores” of submitted predictions as evaluated against a held back portion of the data. And to foster collaboration in the Challenge community, organizers are planning to roll out a few rewards during the Challenge. These will encourage participants to, for example, submit code for their models so that it can be used by others to build new and improved hybrid models (for which both the creator and borrower of code will be rewarded) and to write-up the so-called “provenance” description for their favorite model, describing the analytical steps taken to build that model, so that others can have a better understanding of how different models are constructed. Finally, funds from the DREAM conference sponsors, including the NCI’s Magnet Center (at Columbia University) and IBM Research, will be used to provide small travel grants to top performing teams to present their results at the annual DREAM conference.

“We anticipate that this Challenge will attract a lot of enthusiasm from the modeling community due to the size, scale, and uniqueness of this fantastic dataset,” said Gustavo Stolovitzky, co-founder of the DREAM project and a key leader on the planning of this Challenge. “With the special features in this Challenge, such as the real time leaderboard and incentives to share and borrow model code, which in the 2012 Sage-DREAM Breast Cancer Prognosis Challenge attracted over 1500 models, we expect that the Toxicogenetics Challenge will also elicit submission of thousands of model predictions.”

Three Challenges open today

The NIEHS-NCATS-UNC DREAM Toxicogenetics Challenge is one of three Challenges that Sage Bionetworks and DREAM opened to the public today. The two other Challenges are:

  • The Heritage Provider Network-DREAM Breast Cancer Network Inference Challenge: Infer the signaling networks in breast cancer cell lines
  • The Whole Cell Parameter Estimation Challenge: Infer the kinetic parameters underlying biological processes in whole cell models

More information on these DREAM8 Challenges including how to participate is available here: https://www.synapse.org/ - !Challenges:DREAM8

ABOUT THE UNIVERSITY OF NORTH CAROLINA TEAM AT CHAPEL HILL

Chartered in 1789 as the nation’s first public university, The University of North Carolina at Chapel Hill has earned a reputation as one of the best universities in the world. UNC’s Gillings School of Global Public Health is the top-ranked public school of public health in the nation. The work that has led to this collaboration was conducted by the students and staff of the Carolina Center for Computational Toxicology under direction of Ivan Rusyn, M.D., Ph.D., and Fred Wright, Ph.D., who are faculty members in UNC-Gillings’ Department of Environmental Science and Engineering and Department of Biostatistics.

ABOUT THE DREAM PROJECT: http://www.the-dream-project.org.

ABOUT SAGE BIONETWORKS: http://sagebase.org/.

ABOUT NIEHS

The NIEHS supports research to understand the effects of the environment on human health and is part of NIH. For more information on environmental health topics, visit http://www.niehs.nih.gov. Subscribe to one or more of the NIEHS news lists to stay current on NIEHS news, press releases, grant opportunities, training, events, and publications.

Contacts

Sage Bionetworks
Stephen Friend, 206-667-2101
friend@sagebase.org
or
Gustavo Stolovitzky, 914-945-1292
gustavo@us.ibm.com

Release Summary

Challenges engage diverse communities to competitively solve a problem by placing scientific data, tools and predictive models into an open Commons – in effect, “crowdsourcing” data analysis.

Sharing

Contacts

Sage Bionetworks
Stephen Friend, 206-667-2101
friend@sagebase.org
or
Gustavo Stolovitzky, 914-945-1292
gustavo@us.ibm.com