log Pow Prediction for Cyclic Peptides Using Molecular Simulations
Takuya FUJIE *, Masahito OHUE
School of Computing, Institute of Science Tokyo
The n-octanol/water partition coefficient (log Pow) is widely used as an approximate indicator of membrane permeability, typically determined through experimental measurements. However, such experimental methods often entail significant time and cost, limiting their scalability. To address these limitations, chemoinformatics-based approaches have been developed for predicting log Pow. Among these, atom-based methods such as AlogP (or MolLogP), implemented in software tools (e.g., RDKit), are commonly employed. Nevertheless, these methods struggle to provide accurate predictions for molecules that undergo substantial structural changes in solution or for compounds where specific solute–solvent interactions, such as hydrogen bonding, play a critical role especially in the case of large or flexible molecules. In this study, we propose a computational chemistry-based strategy to calculate log Pow, aiming to improve permeability predictions for medium-sized molecules where conventional chemoinformatics methods fall short. Specifically, we focus on cyclic peptides and perform conformational sampling in both water and n-octanol environments using molecular dynamics (MD) simulations. Subsequently, solvation free energies are computed to estimate log Pow values, incorporating the influence of solution-phase conformational ensembles.