Professor Amir Farimani
Mechanical engineering, Machine Learning and Bio-engineering
Carnegie Mellon University
Pittsburgh, PA

Monday, April 15, 2019
3:00 pm (CT)
3269 Beckman Institute

Abstract

G-Protein Coupled Receptors (GPCRs), comprises over one-third of the targets of all FDA-approved drugs. One such GPCR, the µ Opioid Receptor (µOR), epitomizes the benefits and drawbacks of existing GPCR drugs. Opioid chronic pain medications, such as morphine and hydrocodone, are µOR agonists that achieve their main therapeutic aim of analgesia, yet cause severe side effects, such as respiratory depression and addiction. Conformational changes of these receptors and intermediate states play a key role in biased agonism. To this end, we used hundreds of MD simulation trajectories and machine learning (ML) to elucidate the conformational changes in Opioid receptors during their activation. The signal transduction pathways and ligand-directed conformational changes will be discussed. In the second part, I’ll talk about different ML algorithms used in molecule/material discovery and present their accuracies for different tasks. The limitations of these algorithm along with their interpretability will also be discussed.