SFCi 2025

Modern Virtual Screening workflows

Speaker:
David Papin, Principal Scientist II, Applications Science at Schrödinger

Abstract:
Schrödinger has a long history of developing virtual screening technologies. Modern virtual screening faces new challenges, particularly with the emergence of ultra-large chemical libraries over the past 10 years. As Schoichet et al. highlighted [1], there is a clear need to explore a much larger chemical space to improve the number and quality of hits found. We will be presenting a modern virtual screening workflow that efficiently screens ultralarge libraries. This workflow combines ligand-based approaches and machine learning-guided docking with advanced scoring methods, such as – 1D-sim which measures molecular similarity by projecting 2D structures into a single atomic coordinate [2]. When combined with Shape Screening, it gives rise to a cascaded screening workflow named QuickShape [3]. – GlideWS [4]: an advanced docking method that combines enhanced ligand sampling and a physics-based empirical scoring function to improve hit discovery and pose prediction in virtual screening – ABFEP (Absolute Binding Free Energy Perturbation) [5]: a highly accurate, physics-based computational method that calculates absolute binding free energy We will also emphasize the benefits of screening large libraries with a combination of machine learning and physics-based methods (Active Learning workflows).