High Confidence Predictions of Drug−Drug Interactions: Predicting Affinities for Cytochrome P450 2C9 with Multiple Computational Methods

Matthew G. Hudelson*†, Nikhil S. Ketkar‡, Lawrence B. Holder*‡, Timothy J. Carlson§, Chi-Chi Peng, Benjamin J. Waldher‡ and Jeffrey P. Jones*
Department of Mathematics, Washington State University, Post Office Box 643113, Pullman, Washington 99164-3113, School of Electrical Engineering and Computer Science, Washington State University, Post Office Box 642752, Pullman, Washington 99164-2752, Department of Pharmacokinetics and Drug Metabolism, Amgen, 1120 Veterans Boulevard, South San Francisco, California 94080, and Department of Chemistry, Washington State University, Post Office Box 644630, Pullman, Washington 99164-4630
J. Med. Chem., 2008, 51 (3), pp 648–654
DOI: 10.1021/jm701130z
Publication Date (Web): January 19, 2008
Copyright © 2008 American Chemical Society
* To whom correspondence should be addressed. Tel.: 509-335-5983 (J.P.J., for general methods). Fax: 509-335-8867 (J.P.J.). E-mail: jpj@wsu.edu (J.P.J.); Tel.: 509-335-6138 , Fax: 509-335-3818, E-mail: holder@wsu.edu (L.B.H., for SUBDUE); Tel.: 509-335-3125 , Fax: 509-335-1188, E-mail: mhudelson@wsu.edu (M.G.H., for LWRP, NERP, Gravity).
,
†

Department of Mathematics, Washington State University.

,
‡

School of Electrical Engineering and Computer Science, Washington State University.

,
§

Department of Pharmacokinetics and Drug Metabolism, Amgen.

,

Department of Chemistry, Washington State University.

Abstract

Abstract Image

Four different models are used to predict whether a compound will bind to 2C9 with a Ki value of less than 10 µM. A training set of 276 compounds and a diverse validation set of 50 compounds were used to build and assess each model. The modeling methods are chosen to exploit the differences in how training sets are used to develop the predictive models. Two of the four methods develop partitioning trees based on global descriptions of structure using nine descriptors. A third method uses the same descriptors to develop local descriptions that relate activity to structures with similar descriptor characteristics. The fourth method uses a graph-theoretic approach to predict activity based on molecular structure. When all of these methods agree, the predictive accuracy is 94%. An external validation set of 11 compounds gives a predictive accuracy of 91% when all methods agree.

View: Full Text HTML | Hi-Res PDF | PDF w/ Links

Tools

History

  • Published In Issue February 14, 2008
  • Article ASAPJanuary 19, 2008
  • Received: September 11, 2007

Recommend & Share

Related Content

Other ACS content by these authors: