Recently, docking libraries have expanded from three million “off-the-shelf” to over three billion “make-on-demand” molecules (and in other groups, over 10 billion).  It is easy to imagine that as we expand to ever larger libraries the docking scoring functions, with all their approximations, will get overwhelmed by decoys.  While there is some evidence that that can happen, current evidence suggests that as ever larger libraries are docked, affinities if anything improve. Meanwhile, plots of hit-rates vs. score for the dopamine and recently for the Sigma2 receptors, tested at scale, suggest that docking score remains a strong categorizer even as libraries grow. Here we investigate why that may be, and prospects for looking ahead into ever larger libraries, both by brute-force docking and by ML-based methods.  Case studies seeking to discover new ligands with new pharmacologies for analgesia and psychedelic targets will also be presented.

Ultra-large library docking: Promise and peril

ACS Spring 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.