Revealing Adverse Outcome Pathways from Public High-Throughput Screening Data to Evaluate New Toxicants by a Knowledge-Based Deep Neural Network Approach

• DOI: 10.1021/acs.est.1c02656
• Publication Date: 2021-07-26T04:36:05Z
• Authors: Heather L. Ciallella; Daniel P. Russo; Lauren M. Aleksunes; Fabian A. Grimm; Hao Zhu
• Keywords: Adverse Outcome Pathway

Traditional experimental testing to identify endocrine disruptors that enhance estrogenic signaling relies on expensive and labor-intensive experiments. We sought design a knowledge-based deep neural network (k-DNN) approach reveal organize public high-throughput screening data for compounds with nuclear estrogen receptor α β (ERα ERβ) binding potentials. The target activity was rodent uterotrophic bioactivity driven by ERα/ERβ activations. After training, the resultant successfully inferred critical relationships among bioassays, shown as weights of 6521 edges between 1071 neurons. uses an adverse outcome pathway (AOP) framework mimic initiated ERα endogenous estrogens (i.e., mimetics). k-DNN can predict mimetics activating neurons representing several events in pathway. Therefore, this virtual model, starting from compound's chemistry initiating activation ending bioactivity, efficiently accurately prioritize new (AUC = 0.864-0.927). This method is potential universal computational toxicology strategy utilize characterize hazards potentially toxic compounds.