ToxCast Program
Predicting Hazard, Characterizing Toxicity Pathways, and Prioritizing the Toxicity Testing of Environmental Chemicals
ToxCast Navigation
ToxCast™ Data Analysis Summit
Click here to go to registration and travel information page for the TDAS Meeting
Transforming Toxicity Testing From In Vivo to In Vitro:
A Computational Toxicology Challenge
The First ToxCast™ Data Analysis Summit
Hosted by U.S. EPA"s National Center for Computational Toxicology
EPA Campus, Research Triangle Park NC
May 14-15, 2009
The EPA ToxCast™ Program is developing novel approaches to predict chemical toxicity using high-throughput and high content in vitro assays. ToxCast™ has produced data on 320 chemicals from over 500 in vitro assays, and about 75 in vivo endpoints, providing a powerful dataset for evaluating analysis approaches.
The EPA made the ToxCast™ data available to analysis partners who signed a Materials Transfer Agreement (MTA). We invited our partners to apply their best analysis strategies, and present their findings at the first ToxCast™ Data Analysis Summit (TDAS). The goal is to find robust analysis methods for predicting whole animal or human chemical toxicity from the in vitro data.
The meeting will be held in the Building C auditorium, room C111, on EPA’s campus in Research Triangle Park, NC.
AGENDA
(All times are Eastern Daylight Savings Time (EDT))
Day 1: Thursday, May 14 EPA RTP campus building C auditorium (Room C111) |
||
8:30am-8:40 | Robert Kavlock | Welcoming Remarks and Goals of the Summit (EPA/ORD/NCCT) |
8:40am-9:00 | Jim Jones | Using ToxCast™ for Chemical Screening and Prioritization in the Real World (EPA/OPPTS) |
9:00am-10:10am | Introduction to ToxCast- Phase I |
|
9:00-9:15 | David Dix | Overview of the ToxCast™ Research Program: Applications to Predictive Toxicology and Chemical Prioritization (EPA/ORD/NCCT) |
9:15-9:45 | Keith Houck | Characteristics of the ToxCast™ In Vitro Datasets from Biochemical and Cellular Assays (EPA/ORD/NCCT) |
9:45-10:00 | Matt Martin | Characteristics of the ToxRefDB In Vivo Datasets from Chronic, Reproductive and Developmental Assays (EPA/ORD/NCCT) |
10:00-10:30 | BREAK--LIGHT REFRESHMENTS PROVIDED | |
10:30am-11:30am | Predicting In Vivo Toxicity from ToxCast™ In Vitro Data |
|
10:30-11:00 | Richard Judson | NCCT Predictive Signatures from ToxCast™ (EPA/ORD/NCCT) |
11:00-11:30 | Barry Hardy | Initial OpenTox Evaluation of ToxCast Phase 1 Datasets(OpenTox) |
11:30-12:30 | LUNCH--ON YOUR OWN (there is a café onsite) |
|
12:30pm-4:45pm | Mining ToxCast™ Data |
|
12:30-1:00 | Alex Tropsha (U of North Carolina) | Prediction of animal toxicity endpoints of ToxCast Phase I compounds using a combination of chemical and biological in vitro descriptors |
1:00-1:30 | Lyle Burgoon (Michigan State U) | Biomarker Identification using Graph Theoretic and Particle Swarm Optimization-based Support Vector Machine Analysis of the Phase I ToxCast™ Dataset |
1:30-2:00 | William Welsh (U Med Dentistry NJ) | Biological profile analysis of the ToxCast chemicals: Linking biological activity profiles to molecular structure |
2:00-2:30 | Rusty Thomas (Hamner Institutes) | An Integrated In Vitro and Computational Approach to Define the Exposure-Dose-Toxicity Relationships In High-Throughput Screens |
2:30-3:00 | BREAK-- LIGHT REFRESHMENTS PROVIDED |
|
3:00-3:30 | Weida Tong (FDA/NCTR) | Prediction of liver toxicity in the animal study using the mechanistically relevant in vitro screening assay data – Lessons learned from the EPA ToxCast™ project |
3:30-4:00 | Fred Wright (U North Carolina) | Prediction of in vivo toxicity endpoints from ToxCast Phase I data using a variety of machine learning approaches |
4:00-4:30 | Alison Motsinger (NC State U) | Predictive Modeling of Toxicity Outcomes with Grammatical Evolution Neural Networks |
4:30-4:45 | Open Discussion on ToxCast™ Predictive Modeling, Introduction to Poster Session |
|
5:00pm-7:00pm | Poster Session from submitted abstracts on ToxCast™ Data Analysis Atrium of Building B Light refreshments and beverages provided for meeting registrants. |
|
Day 2: Friday, May 15 EPA RTP campus building C auditorium (Rooms C111) |
||
8:30am-10:50am | Mining ToxCast™ Data for Predictive Toxicity Signatures (cont.) |
|
8:30-8:50 | James Rathman (Ohio State U) | Navigating through chemical and biological data domains of ToxCast™ |
8:50-9:10 | Christodoulos Floudas (Princeton U) | Predicting in vivo toxicities using optimal methods for re-ordering and machine learning |
9:10-9:30 | Nina Jeliazkova (IdeaConsult Ltd.) | Hierarchical Multi-label Classification of ToxCast Datasets |
9:30-9:50 | Jun Huan (U of Kansas) | Learning from Multiple Data Sources and Its Applications in Chemical Toxicity Prediction |
9:50-10:10 | BREAK—LIGHT REFRESHMENTS PROVIDED |
|
10:10-10:30 | Jessie Xia (Nat’l Inst Statistical Sciences) | Hierarchical Predictive Analysis of Chemical Toxicity By Recursive Partitioning Using the Phase-I ToxCast Dataset |
10:30-10:50 | Johann Gasteiger (Molecular Networks GmbH) | Chemical Reactivity in Metabolism and Degradation Reactions in the Risk Assessment Workflow |
10:50am-11:40am | Future Directions for ToxCast™ |
|
10:50-11:10 | Matt Martin (EPA/ORD/NCCT) | Biotransformation and ToxCast™ |
11:10-11:30 | Holly Mortensen (EPA/ORD/NCCT) | Mapping Human Toxicity and Disease Pathways in ToxCast™ |
11:30-12:30 | LUNCH-- ON YOUR OWN (there is a café onsite) |
|
12:30pm-3:00pm | Moving Forward with Research and Applications of ToxCast™ |
|
12:30-1:00 | David Reif (EPA/ORD/NCCT) | Summary of Approaches and Predictions |
1:00-1:45 | Issues Raised by Participants: single slides from volunteers |
|
1:45-2:45 | Robert Kavlock (Moderator) | Facilitated Discussion on ToxCast™ Predictions and Applications
Key Topics: |
2:45-3:00 | Richard Judson (EPA/ORD/NCCT) | Action Items and Closing Remarks |
Overview: The U.S. EPA ToxCast™ Program is developing approaches to predict chemical toxicity using data from high-throughput and high content in vitro assays. Phase I of ToxCast has produced data from 320 chemicals, ~500 in vitro assays and ~100 in vivo endpoints, providing a powerful dataset for evaluating the applicability of various analytic approaches for predicting the potential for an adverse response.
The goal of ToxCast is to develop and verify "toxicity signatures," which are algorithms using in vitro and in silico data to predict in vivo toxicities. These signatures will be used to screen and prioritize chemicals for targeted toxicity testing, and over the next several years EPA would like to screen thousands of compounds. However, successful predictive models will depend on robust and reliable methods that EPA can rely on for making decisions about further testing of environmental chemicals.
This first ToxCast Data Analysis Summit is designed to bring together experts in machine learning, computational chemistry, statistics, high-throughput screening and computational toxicology, with toxicologists and regulatory staff. Plenary talks will describe the ToxCast Program and a series of issues related to toxicity prediction, both from a scientific and regulatory standpoint. Speakers will be selected from abstract submitters to describe algorithmic, computational or systems biology approaches to solving these issues.
To further this aim, we invite interested researchers to submit abstracts and present their analyses, using the ToxCast Phase I dataset, at the First ToxCast Data Analysis SummitData Overview: : The collection of ToxCast Phase I chemicals were chosen because high quality, guideline-based animal toxicity data were available. These chemicals are mostly pesticide active compounds for which we have rat and mouse 2-year chronic/cancer, 2-generation reproductive, and developmental toxicity data. For analysis, we will provide ~100 toxicity endpoints from these study types whose value is a "LEL" or lowest effective level at which the endpoint was observed- these are the values to predict. In addition, we will provide other aggregated endpoints derived from clustering analyses. Analysis groups (or analysis partners) are also free to develop other endpoints to predict from the data that we will provide.
A total of 9 in vitro datasets have been produced, reviewed and cleared for research use. These include biochemical receptor and enzyme assays; and cell-based assays measuring RNA and protein, cytotoxicity, cell growth and morphology changes in a variety of human and rodent cell types. For each assay, we will provide chemical concentration values at which the assay becomes active, e.g. IC50 or LEC (lowest effective concentration). Additionally, we will provide molecular descriptors and physicochemical properties calculated from chemical structure.
Name | Description | Number of Assays |
ACEA | Real-time Cell Electronic Sensing | 7 |
Attagene | Transcription factor assays | 80 |
BioSeek | Cell-based protein level assays | 87 |
Cellumen | Cell imaging assays | 10 |
CellzDirect | Transcription assays | 17 |
Gentronix | Genetic toxicity assay | 1 |
NCGC | Nuclear receptor assays | 21 |
Novascreen | Receptor binding and enzyme inhibition assays | 239 |
Solidus | Phase I /II metabolic enzyme vs. cytotoxicity assays | 4 |
ChemClass | Chemical Class Information | 107 |
PhysChem | Calculated physico-chemical properties | 3 |
ToxRefDB Endpoints | Animal study data from chronic / cancer, multigenerational and developmental toxicity guideline studies | 100 |
Further Information:
- Meeting organizers contact information:
- David Dix, dix.david@epa.gov
- Richard Judson, judson.richard@epa.gov
Organizing Committee:
David Dix | (co-chair, EPA/ORD/NCCT) |
Richard Judson | (co-chair, EPA/ORD/NCCT) |
Lyle Burgoon | (Michigan State University) |
William Mundy | (EPA/ORD/NHEERL) |
Aldert Piersma | (RIVM, The Netherlands) |
Ivan Rusyn | (University of North Carolina at Chapel Hill) |
Rusty Thomas | (The Hamner Institutes for Life Sciences) |
Weida Tong | (FDA/NCTR) |
Alex Tropsha | (University of North Carolina at Chapel Hill) |
William Welsh | (University of Medicine and Dentistry of New Jersey) |
Fred Wright | (University of North Carolina at Chapel Hill) |