Comparative virtual screening and novelty detection for NMDA-GlycineB antagonists

Abstract

Identification of novel compound classes for a drug target is a challenging task for cheminformatics and drug design when considerable research has already been undertaken and many potent lead structures have been identified, which leaves limited unclaimed chemical space for innovation. We validated and successfully applied different state-of-the-art techniques for virtual screening (Bayesian machine learning, automated molecular docking, pharmacophore search, pharmacophore QSAR and shape analysis) of 4.6 million unique and readily available chemical structures to identify promising new and competitive antagonists of the strychnine-insensitive Glycine binding site (Glycine(B) site) of the NMDA receptor. The novelty of the identified virtual hits was assessed by scaffold analysis, putting a strong emphasis on novelty detection. The resulting hits were tested in vitro and several novel, active compounds were identified. While the majority of the computational methods tested were able to partially discriminate actives from structurally similar decoy molecules, the methods differed substantially in their prospective applicability in terms of novelty detection. The results demonstrate that although there is no single best computational method, it is most worthwhile to follow this concept of focused compound library design and screening, as there still can new bioactive compounds be found that possess hitherto unexplored scaffolds and interesting variations of known chemotypes.

Relating the shape of protein binding sites to binding affinity profiles: is there an association?

“…Our findings are in agreement with a recent study where NMDA receptor antagonists were selected from a library of 8.8 million compounds, applying different virtual screening methods i.e. 2D descriptor-based filtering, molecular docking, QSAR pharmacophore hypothesis and 3D shape search [26]. The best positive hits from each approach were further evaluated by in vitro tests.…”

Identification and Validation of Novel Human Pregnane X Receptor Activators among Prescribed Drugs via Ligand-Based Virtual Screening

“…The integrated use of docking and structureactivity relationship models to determine ligand, substrate, or inhibitor specificity has greatly advanced our understanding of the mechanism of receptors and drug transporters (Khandelwal et al, 2008;Krueger et al, 2009). To date, several combinations of structure-and/or ligand-based methods have been reported to characterize hPXR and activator interactions (Gao et al, 2007;Ekins et al, 2008;Khandelwal et al, 2008;Yasuda et al, 2008), but application of this strategy to Previous docking studies on hPXR indicated that directly using the "cutoff score" from docking programs was limited for prediction and classification of hPXR activators and nonactivators (Ekins et al, , 2009Khandelwal et al, 2008;Yasuda et al, 2008).…”

Ionic Channels as Targets for Drug Design: A Review on Computational Methods

“…Krueger et al validated different techniques for virtual screening to identify potential competitive antagonists of the NMDA receptor glycine binding site located in the NR1 subunit [113]. Molecular docking, pharmacophore search, pharmacophore QSAR and Bayesian machine learning were used for virtual screening of 4.6 million chemicals.…”