Inclusion of molecular flexibility in partition coefficient calculations on oligopeptides using stochastic sampling

Fragment-based methods for computing water/1-octanol partition coefficients (log P), which employ two-dimensional structural representations, can often yield estimates of variable quality for flexible structures. We report a novel approach for calculating log P which combines continuum solvation potentials and conformational search algorithms. In initial studies employing linear dipeptides, excellent agreement is obtained between calculated and experimental log P values using either stochastic sampling or systematic search methods to model the conformations adopted by these dipeptides in both solvent phases. Our results are consistent with proposals that the physical basis of certain correction terms used in empirical methods for log P estimation is due to neglect of the redistribution of conformer populations as compounds partition between aqueous and hydrophobic environments.<>