Atomistic LJ parametrization of liquid properties for PC and MTP electrostatics

Preliminary notes

It provides a series of scripts to parametrize Lennard-Jones coefficients for liquid simulations with either point charges (PC) or multipole (MTP) electrostatics. This package requires Gaussian, CHARMM, NAMD, and VMD. It also requires a PDB of the compound of interest, and CHARMM-compatible topology and parameter files. Note that we assume point charges (PC) or multipole (MTP) parameters have already been determined (see Related work below).

Installation

Simply clone the package to your computer

:::bash
git clone git@bitbucket.org:tbereau/lj-fit.pc-mtp.git

Please beware that scripts need to be run from the original source directory, as files rely upon one another. Either call scripts from their original location or create symbolic links (i.e., ln -s file). Do not copy individual scripts away from the directory.

In the following, all references to the directory where the package was copied will be referred to as ~/$DIR.

• Setup config file

Usage

1. Gas-phase properties: To optimize the Lennard-Jones (LJ) parameters of a compound, we first run ab-initio calculations of dimer conformations: compound-water. We then extract all raw energies (i.e., ab initio and force field), and finally we provide the means to fit the LJ coefficients to the ab initio data alone.

   • Ab initio calculations of compound-water dimers
   • Extract raw energies from ab initio and force-field data
   • LJ-coefficient determination
     • Fit LJ coefficients from ab-initio data
     • Monte-Carlo parameter variation

2. Liquid properties: While the afore-mentioned fit does not necessarily provide accurate thermodynamic properties for liquids, we use the ab-initio data to aim the parameter scan for liquid properties. In that sense, the methods presented below rely on the data obtained above. The methods presented here rely on experimental data, namely the pure-liquid density and heat of vaporization (both at the same temperature).

   • Generate liquid PSF and CRD
   • Parametrization
     • Fragment-based
       • Fragment-based parametrization
       • Extract sets close to experimental data from fragments
     • All atom types at once
       • Filter Monte-Carlo data to avoid repeating sets
       • Liquid simulations from Monte-Carlo parameter variation
       • Extract sets close to experimental data from Monte Carlo

3. Misc.: Other tools to parametrize the LJ coefficients of a force field.

   • Rescale nonbonded LJ for MTP in a perturbative spirit

Related work

• Point-charge and multipole parameter optimization from the electrostatic potential
• Compute hydration free energies using CHARMM

References

When using this work, please cite the following papers:

• Bereau, Kramer, Monnard, Nogueira, Ward, Meuwly, Scoring Multipole Electrostatics in Condensed-Phase Atomistic Simulations, J. Phys. Chem. B 2013
• Bereau, Kramer, Meuwly, Leveraging Symmetries of Static Atomic Multipole Electrostatics in Molecular Dynamics Simulations, Submitted