March 23 @ 4:00 – 5:00 pm
Speaker: Alexander Tropsha, PhD
Presented from UNC-CH
Broadcast Link: Seminar
Identifying chemical structures associated with adverse drug reactions (ADRs) such as Stevens Johnson Syndrome (SJS) can focus surveillance, provide insights for drug design, and help with safer prescriptions. Quantitative Structure-Activity Relationship (QSAR) models can predict ADRs, and thus provide early warnings of potential hazards. Using VigiBase, a unique collection of international drug safety data maintained by the Uppsala Monitoring Center, a dataset of 364 drugs were analyzed and were positively or negatively associated with SJS. Chemical descriptors were computed from drug molecular structures and machine learning approaches such as Random Forest and Support Vector Machines were used to develop QSAR models. By analyzing QSAR models for descriptor importance, novel chemical alerts were discovered (substructures) for SJS that afforded fewer false positives than previously known alerts. Requiring chemical structures only, QSAR models provide effective computational means to flag potential harmful drugs for subsequent targeted surveillance and pharmacoepidemiological investigations.
Alexander Tropsha, Ph.D. is a K.H. Lee Distinguished Professor and Associate Dean for Pharmacoinformatics and Data Science at the UNC Eshelman School of Pharmacy at UNC-Chapel Hill. Professor Tropsha obtained his Ph.D. in Chemical Enzymology in 1986 from Moscow State University, Russia. He came to UNC-Chapel Hill in 1989 as a postdoctoral fellow and became faculty in the School of Pharmacy in 1991. His research interests are in the areas of Computer-Assisted Drug Design, Cheminformatics, Structural Bioinformatics, and Computational Toxicology. He has written and co-authored more than 200 peer-reviewed research papers, reviews, and book chapters, and has co-edited two monographs. His research has been supported by multiple grants from the NIH, NSF, EPA, DOD, as well as private companies.