You are here:
Uncertainty Quantification in ToxCast High Throughput Screening
Citation:
Watt, Eric AND R. Judson. Uncertainty Quantification in ToxCast High Throughput Screening. PLOS Computational Biology. Public Library of Science, San Francisco, CA, 13(7):1-23, (2018). https://doi.org/10.1371/journal.pone.0196963
Impact/Purpose:
The U.S. Environmental Protection Agency's ToxCast high throughput screening project is used to measure the toxicity of thousands of chemicals for which we have little to no information about their biological activity. In this work, we introduce a new method for uncertainty quantification in ToxCast data. We explore how unavoidable uncertainties in the data result in uncertainties in concentration-response parameters such as potency and efficacy. These uncertainties are then extended throughout to the analysis and interpretation of results for risk assessment. By quantifying these uncertainties through the analysis stages we increase the confidence in the data interpretation and allow for a more robust risk assessment. We also flag chemicals and assays for manual inspection, removal, and retesting so that data quality and model outputs can be further improved. Ultimately this work will be used to refine and improve confidence in numerous models of toxicity that are developed using ToxCast data.
Description:
High throughput screening (HTS) projects like the U.S. Environmental Protection Agency's ToxCast program are required to address the large and rapidly increasing number of chemicals for which little to no toxicity data are available. Concentration-response parameters such as potency and efficacy are extracted from HTS data using nonlinear regression, and models and analyses built from these parameters are used to predict in vitro and in vivo toxicity of thousands of chemicals. How these bioactivity predictions are impacted by uncertainties that stem from parameter estimation and propagate through models and analyses has not been well explored. While data size and complexity make uncertainty quantification computationally expensive for HTS datasets, continued advancements in computational resources have allowed these computational challenges to be met. This study used nonparametric bootstrap resampling to calculate uncertainties in concentration-response parameters from a variety of HTS assays. Using the ToxCast estrogen receptor model for bioactivity as a case study, we highlight how these uncertainties can be propagated through models to quantify the uncertainty in model outputs. Uncertainty quantification in model outputs was used to identify false positives and false negatives and to determine the distribution of model values around semi-arbitrary activity cutoffs, increasing confidence in model predictions. At the individual chemical-assay level, curves with high variability were flagged for manual inspection or retesting, thereby focusing subject-matter-expert time on results that need further input. This work improves the confidence of predictions made using ToxCast data, increasing the ability to use this data in risk assessment.
