In silico identification of potent protein inhibitors commonly requires prediction of a ligand binding free energy (BFE). Thermodynamics integration (TI) based on molecular dynamics (MD) simulations is a BFE calculation method capable of predicting accurate BFE,but it is computationally expensive and time-consuming. In this work, we developed an efficient automated workflow for identifying compounds with the lowest BFE among thousands of congeneric ligands which requires only hundreds of TI calculations.<br/>We virtually screened over 1.3 billions of commercially available compounds for PLpro activity and identified a focused library of 8×103 compounds with N-[(1R)1(1-aryl)ethyl]arylamide scaffold and its derivatives. Selected ligands were then screened by an iterative procedure, which combines relative BFE calculations via a GPU-accelerated MD TI and a machine learning model trained by using an active learning approach.<br/>We developed an automated workflow utilizing coupled AutoML and Active Learning approach: a computationally low-cost proxy machine learning model (initially trained on a small diverse subset of the library with computed alchemical BFEs) iteratively proposes the next compound (with respect to exploitation/exploration tradeoff) for alchemical BFE calculations and updates itself. The proxy ML model underlying algorithm, molecular featuresations are automatically adjusted on each iteration with respect to training data improving performance. This allowed for finding ligands with the lowest BFEs in the predefined focused library yet performing only a small fraction of computationally-intensive calculations. Our results will contribute to targeting PLpro for the identification of potent inhibitors of clinical value for SARS-CoV-2-specific therapy.

Lead Optimization of SARS-CoV-2 papain-like protease inhibitors with free energy calculations accelerated by AutoML and Active Learning

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

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