Release the Data! New Chemical Data, Workshops, and Challenges

By Matthew T. Martin

Scientist prepares a well-plate for high-throughput screening.

Scientist preparing a well-plate for high-throughput screening.

Ever open that cabinet under the kitchen sink, grab that bright blue bottle of window cleaner and wonder exactly what sort of chemicals are floating around in it? Many of you have at one time or another, and for those of you who have never given it a second thought rest assured that my colleagues and I at EPA are dedicated to identifying and categorizing all of the chemicals we might be exposed to on any given day. However, due the expensive, time-consuming process of traditional testing, which assesses one chemical at a time, only a small fraction of the tens of thousands of chemicals currently in commerce have been adequately assessed for potential human and environmental health risks.

To close this data gap and better evaluate potential health risks, we have worked hard in recent years to accelerate the pace of chemical testing. I am proud to say that we have now completed phase two of the multi-year Toxicity Forecaster (ToxCast) project and are publically releasing ToxCast data on 1,800 chemicals evaluated in over 700 high-throughput screening assays. This is a significant accomplishment that we want to share with the scientific community.

The new data is accessible through the new interactive Chemical Safety for Sustainability (iCSS) Dashboard, a web-based application for users to access and interact with the data freely at their own discretion. Users can select the chemicals and data of interest and then score the information to help inform chemical safety decisions.

As part of the data release, I hope the scientific community will take advantage of this new windfall of data and become involved in the ToxCast project by participating in the Predictive Toxicology Challenges. The first two challenges of the series, available through TopCoder and InnoCentive crowd sourcing technology, will ask the scientific and technology community to develop new algorithms to predict lowest effect levels (LELs) of chemicals using the new ToxCast data. Winners will receive monetary prizes to help fund their own planned research, and their solutions will help us determine innovative ways to use ToxCast data to inform decisions made about the chemical safety.

Also, beginning January 14,we are also hosting several stakeholder outreach workshops and webinars to address potential challenges with data translation, accessibility, and any other troubleshooting issues that might arise during the initial data launch. This is an opportunity for the scientific community to provide input on data usage and offer immediate feedback about the new data and the iCSS dashboard.

About the author: Matthew T. Martin is a research biologist within EPA’s National Center for Computational Toxicology, where he is part of the ToxCast team and leads the CSS task for developing predictive models of toxicity using high-throughput screening data. He also serves as the project lead for developing the new CSS Dashboard Web Application.

Editor's Note: The opinions expressed here are those of the author. They do not reflect EPA policy, endorsement, or action.

Please share this post. However, please don't change the title or the content. If you do make changes, don't attribute the edited title or content to EPA or the author.

Seeing Green (and Predicting It)

By Dustin Renwick

Thick mat of green algae at the shore of a lakeWhen you live near a coast, summer means beaches. A relaxing inland getaway often involves the cool waters of a lake.

Except when the shoreline turns green.

Sometimes a mat of algae clogs fishing lines. Other times, lake foam makes the swimming area appear as if someone poured in a few St. Patrick’s Day green beers.

Those algae serve a natural function in the ecosystem. Yet an icky, slimy scene can ruin plans for a day on the water when conditions – generally warm, stagnant water rich in excess nitrogen, phosphorus,or other nutrient pollution – sparks rapid growth.

Ross Lunetta, EPA research physical scientist, leads a team of research and application scientists who proposed a Pathfinder Innovation Project to validate a new algorithm that uses satellite data for predicting algal blooms in freshwater systems.

Specifically, the team’s project targets cyanobacteria, known to make humans and animals sick with symptoms such as respiratory distress and skin rashes. On the basis of algae cell counts, more than a quarter of lakes nationwide have enough cyanobacteria for moderate to high risk according to the most recent National Lakes Assessment Report in 2009.

Tallying the density of cyanobacteria cells in a water body can provide an estimate of potential exposure risk. But sampling more than a handful of the nation’s lakes can be costly and slow. Plus, current satellite data and its analysis fall short.

Existing field measurement programs were not designed to provide data that researchers can readily use to calibrate and validate satellite-based observations. And satellites can’t discriminate between the sizes or the many species of cyanobacteria, some of which don’t produce toxins.

“It’s not necessarily the same species in Maine as it is in Florida,” Lunetta said. “These things can be very different in size.”

Not knowing the cell volume, which is species specific, makes calculations of blooms cell counts, impacts, and risks a challenge.

The team is working with TopCoder, an online community to bring in outside expertise and innovation through competitions and challenges and expand the search for solutions.

TopCoder represents nearly a half million software developers and algorithm specialists. The company’s process breaks large challenges into small chunks that can be coded, developed, or designed individually. Then TopCoder stacks all the pieces back together into a finished solution.

Imagine a neighborhood full of tinkerers, parts collectors and coding whiz-kids who could gather at your garage to diagnose and fix your car when the “check engine light” flashed. Each person could solve a problem within his or her specialty, and the cohesive result benefits from these specific skills.

The team’s predictive algorithm will take a few more months to design and even longer to validate, but the potential benefits are clear. Water forecasts and public health officials could alert anyone who might consider a day at the lake, and researchers could focus their efforts.

“If you have limited resources, and you can only collect five samples but you have 50 lakes,” Lunetta said, “you can pick the ones the model tells you will most likely become a problem.”

About the author: Dustin Renwick works as part of the innovation team in the EPA Office of Research and Development.

Editor's Note: The opinions expressed here are those of the author. They do not reflect EPA policy, endorsement, or action.

Please share this post. However, please don't change the title or the content. If you do make changes, don't attribute the edited title or content to EPA or the author.