Fig. 3From: Exploring the ability of machine learning-based virtual screening models to identify the functional groups responsible for bindingThe distribution of the number of Hydrogen Bond Acceptor/Donors in each ligand atom for the Polar_ZINC dataset, across the set of actives and decoys. The actives and decoys have almost identical distributions, suggesting that a virtual screening model would not be able to use the number of ligand pharmacophores to predict “binding”Back to article page