Abstract
In silico models were developed for predicting high animal clearance using naïve Bayesian classification and extended connectivity fingerprints. Validation and test sets were created from a structurally diverse database of mouse, rat, dog, and monkey clearance (CL) representing approximately 20 000 unique compounds. Model performance was compared with experimental predictors used widely in drug discovery, namely in vitro intrinsic clearance (CLi) and CL from a lower preclinical species.
The Bayesian model for dog CL was a better predictor than experimental rat or mouse CL. The Bayesian model for rat CL performed at least as well as mouse CL. Bayesian models outperformed mouse, rat, and monkey CLi for predicting mouse, rat, and monkey CL, respectively.
These models can be used to optimize chemical libraries, direct new chemical synthesis and increase efficiency of screening cascades for lead optimization while reducing overall drug discovery cost, time and animal usage.
Acknowledgements
The authors wish to thank past and present GSK Drug Discovery scientists who synthesized and characterized the biological properties of the molecules herein studied. We would also like to thank Amber Anderson (GSK) and Robert Gagnon (GSK) for expert statistical input and John Conway (Accelrys) and Keith Ward (Bausch & Lomb) for critical review of the manuscript.
Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.