The
test on small molecule free energies of transfer from a vacuum to
water was done on a data set of 504 neutral organic small molecules51 (link) taken from David Mobley’s group. The
solvation energies of all of these molecules have been experimentally
determined, with the range from −11.95 to 3.16 kcal/mol.
Solvation energy Gsol has two components,
polar and nonpolar: where Gpolar indicates
the polar (electrostatic) term and Gnonpolar denotes the nonpolar term.
The polar component of solvation
energy was calculated as the grid
energy difference of the system in water and in a vacuum:
The above grid energies were calculated keeping the corresponding
small molecule at the same grid position to cancel the grid artifacts.
Specific considerations were made for the calculations in a vacuum
since one has to define the molecular surface in this case (the border
between molecule and vacuum). Note that in our approach the dielectric
function is continuous and runs throughout the entire space and is
designed to describe dielectric properties of the molecule in water.
Here, we assume that the properties of molecules are unchanged as
they are moved from water to a vacuum. Thus, following the strategy
implemented in ZAP,40 the molecular surface
of molecules is defined by applying a specific cutoff for the dielectric
constant, εcutoff. The cutoff was varied in the protocol
to obtain the best fit against experimental data.
The nonpolar
term of solvation energy Gnonpolar is
calculated via the accessible surface area method:52 (link) where γ
and b are constants
and SA denotes the solvent accessible surface area, which is calculated
using Naccess2.1.1 (http://www.bioinf.manchester.ac.uk/naccess/ ).
The force field used in the calculations was AM1-BCC,53 which is part of general AMBER force field (GAFF).50 (link) In order to optimize the parameters, reference
εin was varied from 0.1 to 4.0, the value of normalized
variance σi was varied from 0.80
to 1.40, and the value of epsilon cutoff εbnd was
varied from 4.0 to 60.0. For each combination of the σi and εbnd values, the least-squares
method was used to obtain the optimized γ and b constants. The best parameters are shown in theResults section (note that because of the different nature
of the process, these values are not expected to be the same as those
obtained in pKa calculations. SeeDiscussion for details).
test on small molecule free energies of transfer from a vacuum to
water was done on a data set of 504 neutral organic small molecules51 (link) taken from David Mobley’s group. The
solvation energies of all of these molecules have been experimentally
determined, with the range from −11.95 to 3.16 kcal/mol.
Solvation energy Gsol has two components,
polar and nonpolar: where Gpolar indicates
the polar (electrostatic) term and Gnonpolar denotes the nonpolar term.
The polar component of solvation
energy was calculated as the grid
energy difference of the system in water and in a vacuum:
The above grid energies were calculated keeping the corresponding
small molecule at the same grid position to cancel the grid artifacts.
Specific considerations were made for the calculations in a vacuum
since one has to define the molecular surface in this case (the border
between molecule and vacuum). Note that in our approach the dielectric
function is continuous and runs throughout the entire space and is
designed to describe dielectric properties of the molecule in water.
Here, we assume that the properties of molecules are unchanged as
they are moved from water to a vacuum. Thus, following the strategy
implemented in ZAP,40 the molecular surface
of molecules is defined by applying a specific cutoff for the dielectric
constant, εcutoff. The cutoff was varied in the protocol
to obtain the best fit against experimental data.
The nonpolar
term of solvation energy Gnonpolar is
calculated via the accessible surface area method:52 (link) where γ
and b are constants
and SA denotes the solvent accessible surface area, which is calculated
using Naccess2.1.1 (
The force field used in the calculations was AM1-BCC,53 which is part of general AMBER force field (GAFF).50 (link) In order to optimize the parameters, reference
εin was varied from 0.1 to 4.0, the value of normalized
variance σi was varied from 0.80
to 1.40, and the value of epsilon cutoff εbnd was
varied from 4.0 to 60.0. For each combination of the σi and εbnd values, the least-squares
method was used to obtain the optimized γ and b constants. The best parameters are shown in the
of the process, these values are not expected to be the same as those
obtained in pKa calculations. See
