single-solute

Practices for a single-molecule simulation of isolate solute

The solvation free energy is the free-energy change for turning on the solute-solvent interaction. The potential parameters are to be kept unchanged when the solvation free energy is obtained. The simulation conditions should be the same between the solution and reference solvent. When a lattice-sum method (Ewald, PME, or PPPM) is used, for example, all the parameters including the real-space cutoff and the number of reciprocal meshes are identical between the two systems. The Lennard-Jones part is also treated in parallel, and the scheme of potential truncation, the cutoff, and the switching range are the same.

When the solute molecule of interest is flexible, its simulation at isolated state is to be done prior to the erdst run for the reference solvent. To keep the invariance of potential parameters in computation of ensemble averages, the following practices are recommended in the single-molecule simulation of an isolated solute.

Option A: No use of a lattice-sum method (PME or PPPM)

1) The simulation is to be done without periodic boundary condition. Or if the periodic boundary condition must be used in user's MD software, the simulation is done in an ensemble with fixed volume (like NVT, not NPT) and the volume is much larger than the size of the solute molecule.
2) The center of mass of the solute is fixed (at the origin). This is to avoid an unlimited growth of the coordinate values of the single solute. When the center of mass is not fixed in the single-solute simulation, Error in erdst: The minimum of the energy coordinate is too large may appear in the erdst run for the reference solvent.
3) The electrostatic interaction is set to its bare form of 1/r (not Ewald, PME, or PPPM), without a cutoff (or with the cutoff distance set essentially to infinity)..
4) The Lennard-Jones interaction is treated in the same way as in the solution and reference solvent; the cutoff scheme and parameters of LJ should be the same among the isolated solute, solution, and reference solvent
5) If user's MD software requires that the cutoff be the same between the electrostatic and Lennard-Jones interactions, no cutoff is used also for the Lennard-Jone part.

With the above 3), a correction is necessary for the electrostatic interaction of the solute when it is transferred to the condensed system. The correction is the self-energy described in the section of Running slvfe: Getting Final output in Quick Start Guide and is performed with wgtslf = 'yes' in parameters_er and slfslt = 'yes' in parameters_fe (see Parameter files for erdst and Parameter files for slvfe).

4) and 5) are actually the same as each other when the solute molecule is small enough and the largest atom-atom distance within the solute is smaller than the Lennard-Jones cutoff distance in 4).

When the solute is not small and the cutoff is required to be the same between the electrostatic and Lennard-Jones interactions in user's MD software, the single-solute run needs to be done with 5). In this case, the difference in the Lennard-Jones cutoff is corrected in erdst by reweighting. This functionality is available as of ver 1.0.5 and can be used with ver 1.0.4 by adopting engmain.F90, engproc.F90, realcal.F90, setconf.F90, and sfecorrect.F90 from trunk.

Option B: Use of a lattice-sum method (PME or PPPM) as in the solution and reference solvent

1) When the ensemble is NVT for the solution and reference solvent, the single-solute simulation can be done by using the periodic cell of the same geometry as those for the solution and reference solvent.
2) When the ensemble is NPT for the solution and reference solvent, the geometry of the periodic cell for the single-solute simulation is set to the average one for the reference solvent.
3) In both cases of 1) and 2), the single-solute simulation is conducted in the NVT ensemble. If the ensemble is NPT for the single-solute run, the cell size may become comparable to the size of a single solute, that is much smaller than those for the solution and reference solvent. In this case, the PME or PPPM parameters can be quite different from those for the solution and reference solvent. Thus, the single solute needs to be simulated in the NVT ensemble by using the cell corresponding to that of the reference solvent.
4) The simulation parameteres for the single-solute run are identical to those for the solution and reference-solvent runs.

Option B can be handled in erdst by setting solute_LatticeSum = 1 in parameters_er. This is available as of ver 1.0.5 and can be used with ver 1.0.4 by adopting engmain.F90, engproc.F90, realcal.F90, setconf.F90, and sfecorrect.F90 from trunk.