This paper presents a large-scale evaluation of the predictive power of ligand-based reverse screening for inferring macromolecular targets of small molecule drugs. Using a machine-learning model trained on ChEMBL data and tested on a Reaxys dataset of over 300,000 active small molecules, the study demonstrates that the model can predict the correct targets for over 51% of the external molecules. The findings highlight the method's usefulness in various drug discovery applications and emphasize the importance of using large, high-quality, non-overlapping datasets for benchmarking such approaches.
Publisher
Communications Chemistry
Published On
May 09, 2024
Authors
Antoine Daina, Vincent Zoete
Tags
ligand-based reverse screening
machine-learning model
predictive power
drug discovery
macromolecular targets
ChEMBL
Reaxys dataset
Related Publications
Explore these studies to deepen your understanding of the subject.
