You are here:
Predicting Chemical-Assay Interference Using Tox21 HTS Data
Citation:
Borrel, A., R. Huang, M. Xia, A. Simeonov, R. Judson, K. Houck, AND N. Kleinstreuer. Predicting Chemical-Assay Interference Using Tox21 HTS Data. Presented at Society of Toxicology annual meeting, Baltimore, MD, March 10 - 14, 2019.
Impact/Purpose:
Abstract for presentation at Society of Toxicology annual meeting March 2019
Description:
The federal Tox21 consortium produces high-throughput screening (HTS) data on thousands of chemicals across a wide range of assays covering critical biological targets. Many of these assays, and those used in other in vitro screening programs, rely on luciferase and fluorescence-based readouts which can be susceptible to signal interference by certain chemical structures resulting in false positive outcomes. Included in the Tox21 portfolio are assays specifically designed to measure interference in the form of luciferase inhibition and autofluorescence via multiple wavelengths (red, blue, and green) and under various conditions (cell-free and cell-based, two cell types). Out of 8,305 unique chemicals tested in the Tox21 interference assays, percent actives ranged from 0.5% (red autofluorescence) to 9.9% (luciferase inhibition) after filtering for curve class, efficacy, and cytotoxicity cutoffs. Bimodal potency distributions were observed among active chemicals, potentially corresponding to specific and non-specific activity. Self-organizing maps and hierarchical clustering were used to relate chemical structural clusters to interference activity profiles. Multiple machine learning algorithms were applied to predict assay interference based on molecular descriptors and chemical properties. The best performing predictive models (accuracies of ~80%) have been included in a web-based tool that will allow users to predict the likelihood of assay interference for any new chemical structure. This abstract does not reflect official EPA or NIH policy.