You are here:
ToxCast Pipeline, Example, and Building Additional Context for Use
Citation:
Paul-Friedman, K. ToxCast Pipeline, Example, and Building Additional Context for Use. Presented at SETAC High-throughput screening and environmental risk assessment, Durham, North Carolina, April 16 - 18, 2018. https://doi.org/10.23645/epacomptox.7098746
Impact/Purpose:
The purpose of this presentation is to orient the toxicology community to the ToxCast data available in invitrodb via the ToxCast Data Pipeline, tcpl. Further, I will introduce ongoing work to improve this pipeline to include uncertainty analysis to provide more context for the use of ToxCast data in various applications. Finally, I will review a timely example of ToxCast data, the high-throughput H295R assay; these assay data are organized in invitrodb, but a custom analysis was performed to create a fit-for-purpose prioritization tool for endocrine bioactivity screening applications. This demonstrates the idea that tcpl is potentially an important module within a broader data analysis workflow.
Description:
The US Environmental Protection Agency (EPA) ToxCast data pipeline (tcpl) has been applied to >1000 assay endpoints as first tier data processing of bioactivity data from high-throughput screening (HTS). The objective of using tcpl, and required database structure, is to organize data from many sources and formats to create a transferrable resource and shared vocabulary for HTS applications. First, tcpl and the structure of invitrodb will be reviewed with a discussion of data “levels,” or formatted data for preparation and completion of curve-fitting. Curve-fitting analysis (levels 4-6) generates concentration-response parameters, including the 50% activity concentration (AC50). Often the data available in level 5, including the AC50 or positive hitcall, may be useful in addressing toxicological questions. However, there are multiple sources of potential variability in these AC50s, resulting from biological variance, experimental error, or curve-fitting procedures. Ongoing work to implement a bootstrap resampling method, available as the toxboot R package, for generation of uncertainty information for AC50s from tcpl will be briefly discussed as a vision of what to expect in future data releases. Such uncertainty information may be useful in interpreting ToxCast data for various uses, e.g. supporting biological key event relationships or for use in preliminary screening level risk assessment. Briefly, toxboot uses smooth nonparametric bootstrap resampling to add random normally distributed noise to give a resampled set of concentration-response values. Useful outputs from toxboot may include hit percent, the probability of a positive hitcall given the collection of 1000 data resamples, and the AC50 median and confidence interval used to quantify AC50 variability. Application of toxboot to ToxCast data provides a robust estimation of curve-fit reproducibility and AC50 uncertainty. Finally, an example of ToxCast data in action will be briefly reviewed: the high-throughput H295R assay, which demonstrates a dataset analyzed with tcpl, as well as a custom analysis to generate fit-for-purpose endocrine bioactivity screening information. This abstract does not necessarily reflect U.S. EPA policy.
URLs/Downloads:
DOI: ToxCast Pipeline, Example, and Building Additional Context for Use
SETAC_HT_TOXCAST_CEETOX_UNCERTAINTY_APR18.PDF (PDF, NA pp, 2519.719 KB, about PDF)