Predicting lipophilicity of drug-discovery molecules using Gaussian process models

ChemMedChem

Volumn 2, Issue 9, 2007, Pages 1265-1267

  Schroeter, Timon S ^a,b   Schwaighofer, Anton ^a   Mika, Sebastian ^c   Ter Laak, Antonius ^d   Suelzle, Detlev ^d   Ganzer, Ursula ^d   Heinrich, Nikolaus ^d   Müller, Klaus Robert ^a,b  

^a FRAUNHOFER FOKUS   (Germany)

^b TECHNISCHE UNIVERSITÄT BERLIN   (Germany)

^c idalab GmbH   (Germany)

^d Research Laboratories   (Germany)

Author keywords

Domain of applicability; Drug design; Gaussian process; Lipophilicity; Machine learning

Indexed keywords

ARTICLE; DRUG DESIGN; THEORETICAL MODEL;

DRUG DESIGN; MODELS, THEORETICAL;

EID: 44449148610     PISSN: 18607179     EISSN: 18607187     Source Type: Journal    
DOI: 10.1002/cmdc.200700041     Document Type: Article

References (15)

1. P. Bruneau, N. R. McElroy, J. Chem. Inf. Model. 2006, 46, 1379-1387.
2. a) A. Schwaighofer, T. S. Schroeter, S. Mika, J. Laub, A. Ter Laak, D. Suelzle, U. Ganzer, N. Heinrich, K.-R. Müller, J. Chem. Inf. Model. 2007, 47, 407- 424;
3. b) K.-R. Müller, S. Mika, G. Rätsch, K. Tsuda, B. Schölkopf, IEEE Trans. Neural Networks 2001, 12, 181-201;
4. c) C. E. Rasmussen, C. K. Williams, Gaussian Processes for Machine Learning, MIT Press, Cambridge MA, 2005.
5. a) G. Orr, K.-R. Müller, Neural Networks: Tricks of the Trade, LNCS, Springer, Berlin, 1998;
6. b) C. M. Bishop, Neural Networks for Pattern Recognition, Oxford University Press, Oxford, 1995.
7. K.-R. Müller, G. RRtsch, S. Sonnenburg, S. Mika, M. Grimm, N. Heinrich, J. Chem. Inf. Model. 2005, 45, 249-253.
8. a) F. R. Burden, J. Chem. Inf. Comput. Sci. 2001, 41, 830-835;
9. b) D. P. Enot, R. Gautier, J. Le Marouille, SAR QSAR Environ. Res. 2001, 12, 461-469;
10. c) P. Tino, I. Nabney, B. S. Williams, J. Lösel, Y Sun, J. Chem. Inf. Comput. Sci. 2004, 44, 1647-1653.
11. J. Quionero-Candela, C. E. Rasmussen, J. Machine Learning Res. 2005, 6, 1939-1959.
12. J. Sadowski, C. H. Schwab, J. Gasteiger, Corina v.3.1, Molecular Networks GmbH Computerchemie, Erlangen (Germany), 2005.
13. R. Todeschini, V. Consonni, Handbook of Molecular Descriptors, Wiley, New York, 2000.
14. To speed up model training and decrease the memory demand, we employed a wrapper script to perform a k-means clustering based on descriptors and to train one GP model for each cluster of up to 5000 compounds. When applying this model, the wrapper considers each GP and chooses the prediction with the highest confidence (smallest predicted error bar).
15. P. Bruneau, J. Chem. Inf. Comput. Sci. 2001, 41, 1605-1616.