Bio: Dr. Zhu is a Professor of Biomedical Informatics and Genomics at the Tulane University Medical School. His major research interest is to use cheminformatics tools to develop predictive models for chemical risk assessments. His current research interests also include data-driven modeling, artificial intelligence algorithm development and computer-aided nanomedicine design. He is the Principal Investigator (PI) of several prestigious research grants (NIH R01, U01, R15, NSF, and etc) with total amount over 10 million dollars. Dr. Zhu is author/co-author of near 100 peer-reviewed journal articles and 10 book chapters with over 8300 citations (H-index as 47). His research was recognized with different awards, such as Rutgers Chancellor’s Award for Outstanding Research and Creative Activity, Society of Toxicology Best Paper Award (three times, 2021, 2023 and 2024), National Institute of Environmental Health Sciences (NIEHS) Extramural Paper of the Month (three times, 2019, 2020 and 2022) and Drug Discovery Today top citation paper of the year (2018).

Abstract: High-throughput screening (HTS) programs generate abundant data on numerous chemicals, elucidating hidden toxicity mechanisms and advancing understanding of adverse outcome pathways (AOPs) in chemical toxicity. However, efficiently organizing and interpreting these data for modeling purposes remains challenging due to their unstructured nature in large repositories like PubChem, variability in HTS program objectives, complexity of assay targets, and inconsistencies in testing protocols, such as varying chemical dosages.

To address these challenges, we developed a mechanistic modeling approach to systematically organize and interpret concentration-dependent HTS data. Using curated datasets, this approach integrated PubChem assay metadata with biological pathway information from WikiPathways, categorizing concentration-response assay outputs within a hierarchical biological pathway framework for AOP model construction. The resultant models generated toxicity scores representing the chemical potency across HTS protein targets within biological pathways. Several pathway results correlated with continuous toxicity endpoints (e.g., acute, maternal, and developmental toxicities) and binary outcomes like hepatotoxicity. Our modeling approach successfully used HTS metadata to create AOP models, elucidating hidden toxicity mechanisms. This computational study offers a novel strategy broadly applicable to risk assessment and drug development processes using extensive public big data.