1:15 Navigating Degrader Property Space with Machine Learning

Time: 9:15 am
day: Day Two Track A PM


  • Degraders occupy distinct chemical space from historical small molecule drugs, leading to poor predictive power in property space from commonly used rules and intuition
  • There has been an expansion in BRO5 models in the literature and from commercial vendors, but internal evaluation indicates Nurix degraders are out of the domain of applicability
  • Machine Learning approaches offer a way to re-learn the rules for relating structure to physical and biological properties for degraders, leading to more efficient computer-directed optimization as well as elucidating human interpretable guidelines for degrader design