Evaluating the Predictivity of Virtual Screening for Abl Kinase Inhibitors to Hinder Drug Resistance
Article first published online: 30 SEP 2013
© 2013 John Wiley & Sons A/S
This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.
Chemical Biology & Drug Design
Volume 82, Issue 5, pages 506–519, November 2013
How to Cite
Gani, O. A. B. S. M., Narayanan, D. and Engh, R. A. (2013), Evaluating the Predictivity of Virtual Screening for Abl Kinase Inhibitors to Hinder Drug Resistance. Chemical Biology & Drug Design, 82: 506–519. doi: 10.1111/cbdd.12170
- Issue published online: 16 OCT 2013
- Article first published online: 30 SEP 2013
- Accepted manuscript online: 8 JUN 2013 07:12AM EST
- Manuscript Accepted: 5 JUN 2013
- Manuscript Revised: 29 MAY 2013
- Manuscript Received: 6 MAR 2013
- virtual screening
Virtual screening methods are now widely used in early stages of drug discovery, aiming to rank potential inhibitors. However, any practical ligand set (of active or inactive compounds) chosen for deriving new virtual screening approaches cannot fully represent all relevant chemical space for potential new compounds. In this study, we have taken a retrospective approach to evaluate virtual screening methods for the leukemia target kinase ABL1 and its drug-resistant mutant ABL1-T315I. ‘Dual active’ inhibitors against both targets were grouped together with inactive ligands chosen from different decoy sets and tested with virtual screening approaches with and without explicit use of target structures (docking). We show how various scoring functions and choice of inactive ligand sets influence overall and early enrichment of the libraries. Although ligand-based methods, for example principal component analyses of chemical properties, can distinguish some decoy sets from active compounds, the addition of target structural information via docking improves enrichment, and explicit consideration of multiple target conformations (i.e. types I and II) achieves best enrichment of active versus inactive ligands, even without assuming knowledge of the binding mode. We believe that this study can be extended to other therapeutically important kinases in prospective virtual screening studies.