The accurate prediction of protein-ligandbinding free energies is a primary objective in computer-aided drug design. Thesolvation free energy of a small molecule provides a surrogate to thedesolvation of the ligand in the thermodynamic process of protein-ligand binding.Here, we use an explicit solvent MD/FEP approachto predict the absolute solvation free energies on a set of 239 small molecules spanning diverse chemical functional groups commonlyfound in drugs and drug-like molecules. We also compare the performance ofabsolute solvation free energies obtained using the OPLS_2005 force field withtwo other commonly used small molecule force fields - General Amber Force Field(GAFF) with AM1-BCC charges and CHARMm-MSIwith CHelpG charges. We find that OPLS_2005 produces the highest correlationwith experimental data (R2=0.94) and the absolute unsigned errorsfor majority of the functional groups are smaller than for AM1-BCC/GAFFor CHelpG/CHARMm-MSI. However, OPLS_2005 performs poorly for certain classesof polar compounds and we show that predictions on these compound classes canbe improved by using a semi-empirical charge assignment method with an implicitbond charge correction (CM1A-BCC).

Prediction of absolute solvation free energies using
free energy perturbation

239th National Meeting (2010)

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.

Prediction of absolute solvation free energies using free energy perturbation