Advertisement

Molecular modeling annual

, Volume 3, Issue 3, pp 142–155 | Cite as

Prediction of the n-Octanol/Water Partition Coefficient, logP, Using a Combination of Semiempirical MO-Calculations and a Neural Network

  • Andreas Breindl
  • Bernd Beck
  • Timothy Clark
  • Robert C. Glen
ORIGINAL PAPER

Abstract

A back-propagation artificial neural net has been trained to estimate logP values of a large range of organic molecules from the results of AM1 and PM3 semiempirical MO calculations. The input descriptors include molecular properties such as electrostatic potentials, total dipole moments, mean polarizabilities, surfaces, volumes and charges derived from semiempirical calculated gas phase geometries. These properties can be related to the molecule′s solubility in hydrophilic or lipophilic media. The input descriptors were selected with the help of a multiple linear regression analysis. The resulting net estimates the logP values of 105 organic compounds with a standard deviation of 0.53 units from the experimental logP values for AM1 and 0.67 units in the case of PM3.

Keywords: Partition coefficient logP AM1 PM3 QSAR neural netIntroduction 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

Supplementary material

In order to view the Electronic Supplementary Material, please download the following zip-file

supp_mat.zip (87 kb)
supp_mat.zip

Copyright information

© Springer-Verlag Berlin Heidelberg 1997

Authors and Affiliations

  • Andreas Breindl
    • 1
  • Bernd Beck
    • 1
  • Timothy Clark
    • 1
  • Robert C. Glen
    • 2
  1. 1.Computer Chemie Centrum des Instituts für Organische Chemie der Friedrich-Alexander-Universität Erlangen-Nürnberg, Nägelsbachstraße 25, D-91052 Erlangen, Germany (clark@organik.uni-erlangen.de).DE
  2. 2.Wellcome Research Laboratories, Langley Park, Beckenham, Kent, BR3 3BS, U. K. Present address: Tripos Inc., 1699 South Hanley Road, St. Louis, MO 63144, USA,US

Personalised recommendations