Date: 29 Oct 2020<\/p>\n
Graduate Student, Zhu Group<\/p>\n
Title: Predictive Modeling of the ER\u03b1\/ER\u03b2 Adverse Outcome Pathway<\/p>\n
Abstract: As defined by the World Health Organization, an endocrine disruptor is an exogenous substance or mixture that alters function(s) of the endocrine system and consequently causes adverse health effects in an intact organism, its progeny, or (sub)populations. Traditional experimental testing regimens to identify toxicants that induce endocrine disruption, including disruptions of the classical genomic ER\u03b1\/ER\u03b2 signaling pathway, can be expensive and time-consuming. Computational modeling has emerged as a promising and cost-effective alternative method for screening and prioritizing potentially endocrine-active chemicals. Modern computational toxicology has moved into a big data era with an enormous amount of data daily generated for chemicals of interest (e.g., drug molecules). Efficiently using this data landscape to quickly and interpretably predict potential toxicants is in high demand. We sought to apply classic machine-learning algorithms, and deep-learning approaches to a panel of over 7500 compounds tested against 18 ER\u03b1\/ER\u03b2 assays in the Environmental Protection Agency\u2019s Toxicity Forecaster high-throughput screening program. The 273 models showed reliable predictivity for new chemicals that have known ER\u03b1\/ER\u03b2 agonist, antagonist, or binding activities but did not undergo experimental testing in these 18 assays. Predictions from this modeling suite can form the basis for future mechanism-driven approaches that combat the \u201cblack box\u201d stereotype of many machine learning algorithms.<\/p>\n","protected":false},"excerpt":{"rendered":"