Premium
Accurate and efficient generalized born model based on solvent accessibility: Derivation and application for Log P octanol/water prediction and flexible peptide docking
Author(s) -
Totrov Maxim
Publication year - 2004
Publication title -
journal of computational chemistry
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.907
H-Index - 188
eISSN - 1096-987X
pISSN - 0192-8651
DOI - 10.1002/jcc.10392
Subject(s) - solvation , solvent models , octanol , chemistry , monte carlo method , computational chemistry , water model , implicit solvation , molecular dynamics , molecule , partition coefficient , thermodynamics , physics , mathematics , organic chemistry , statistics
A novel method for fast and accurate evaluation of the generalized Born radii in macromolecular solvation electrostatics calculations is proposed, based on the solvent accessibility of the first two solvation layers around an atom. The reverse generalized Born radii calculated by the method have correlation coefficient of 98.7% and RMSD of 0.031 Å −1 with the values obtained using a precise but significantly slower numerical boundary element solution. The method is applied to derive an estimate of the free solvation energy difference between octanol and water and to predict Log P octanol–water. A nine‐parameter model is optimized on an 81 compound training set and applied to predict Log P ow for an external evaluation set of 19 drug molecules with RMSD of 0.9. The new GB approximation is also tested in Monte Carlo docking simulations of the fully flexible p53 peptide fragment to MDM2. The best energy solution found in the simulations has RMSD of 2.8 Å to the X‐ray structure. © 2004 Wiley Periodicals, Inc. J Comput Chem 25: 609–619, 2004