fftool

Agilio Padua

This is a Python tool to build force field input files for molecular dynamics.

Contents

• fftool: builds a simulation box and the corresponding force field for systems containing molecules, ions or extended materials. It requires the Packmol software to generate coordinates in the box. The output are files in formats suitable for the LAMMPS, DL_POLY or GROMACS molecular dynamics packages.

• tools/: utility scripts.

• examples/: examples of molecule files and force field databases.

Requirements

• Python

• Packmol to pack molecules and materials in the simultion box.

• PyPy (optional) can bring speed improvements for large systems.

Obtaining

Download the files or clone the repository:

git clone https://github.com/agiliopadua/fftool.git

Tutorial

These are instructions on how to build an initial configuration for a system composed of molecules, ions or materials.

1. For each molecule, ion or fragment of a material prepare a file with atomic coordinates and eventually connectivity (covalent bonds). The formats accepted by this tool are .zmat, .mol, .pdb or .xyz, which are common formats in computational chemistry.

   A .zmat file has the molecule name in the first line, followed by one empy line, then the z-matrix. See the examples directory and the Wikipedia entry for "Z-matrix (chemistry)". Variables can be used in place of distances, angles and dihedrals. Connectivity is inferred from the z-matrix by default. In this case cyclic molecules require additional connect records to close rings. Improper dihedrals can be indicated by improper records. If a reconnect record is present, then connectivity will be guessed based on bond distances from the force field (see below). After the z-matrix and the informations above, the name of a file with force field parameters can be supplied.

   The MDL Molfile .mol file format contains a table with coordinates and also bonds. The name of a file with force field parameters can be given in the first line after the molecule name or in the third line. If the keyword reconnect is present after the force field filename, then connectivity will be deduced based on bond distances from the force field.

   The PDB file format .pdb is widely used for proteins. The name of a file with force field parameters can be given on a COMPND record after the molecule name. Connectivity is deduced from the bond lengths in the force field (CONNECT records are not read).

   The XYZ file format .xyz contains atomic coordinates only. The name of a file with force field parameters can be given in the second line after the molecule name and in this case connectivity is deduced from the bond lengths in the force field.

   There are many tools (Open Babel, Avogadro, VESTA) to create MDL mol files, xyz files or z-matrices. Manual editing of the files is usually necessary in order to match the atom names with those of the force field.

2. Use the fftool script to create an input file for packmol, which will use new _pack.xyz files with atomic coordinates for the components of your system. For help type fftool -h. For example, to build a simulation box with 40 ethanol and 300 water molecules and a density of 38.0 mol/L do:

    fftool 40 ethanol.zmat 300 spce.zmat -r 38.0
   

   Alternatively the side length of the the simulation box (here cubic) can be supplied in angstroms:

    fftool 40 ethanol.zmat 300 spce.zmat -b 20.0
   

3. Use packmol with the pack.inp file just created to generate the atomic coordinates in the simulation box:

    packmol < pack.inp
   

   Difficult convergence may indicate that density is too high, so adjust density or box size if necessary. For more complex spatial arrangements of molecules and materials you can modify the pack.inp to suit your needs (see the Packmol documentation). Atomic coordinates for the full system will be written to simbox.xyz.

4. Use fftool to build the input files for LAMMPS (-l), DL_POLY (-d) or GROMACS (-g) containing the force field parameters and the coordinates of all the atoms (from simbox.xyz):

    fftool 40 ethanol.zmat 300 spce.zmat -r 38.0 -l
   

   If no force field information was given explicitly in the molecule files, a default LJ potential with parameters zeroed will be assigned to atoms. No terms for bonds, angles or torsions will be created. This is suitable when working with non-additive, bond-order or other potentials often used for materials. The input files for MD simulations will have to be edited manually to include an interaction potential for the material.

Deducing Bonds and Angles

When inferring connectivity from interatomic distances, distances in the coordinates file are compared with equilibrium distances specified for bonds in the force field and a tolerance of +/-0.25 angstrom is used to decide if a bond should be present or not. So, the bond lengths in the conformation used as input must be sufficiently close to those in the force field specification for those bonds to be included in the potential energy fonction of the system.

Angles will be assigned to groups of three atoms i-j-k, with i-j and j-k bonded, if the value of the angle in the conformation used as input is within +/-15 degrees of the equilibrium angle in the force field specification. If not, even if the atoms i-j-k are bonded, their angle will not be present in the final potential energy function, although topologically the angle is there. When running fftool to create a force field file (with -l, -d or -g option) a warning message will show which such topological angles have been "removed" because they deviate too much from the equilibrium angles in the force field. This removal of angles avoids problems with atoms that have more than four ligands, such as S or P atoms with five or six ligands. Around these centers there are topological angles of 180 degrees to which no potential energy of bending is attributed in force fields. For example, in the octahedral PF6- anion there are two different values of F-P-F angles: twelve 90 degree angles between adjacent F atoms, and three 180 degree angles between opposite F atoms; only the twelve 90 degree angles contribute with a harmonic potential energy function in most force fields.

The tolerances for bond distances and angle values, 0.25 angstrom and 15 degrees, respectively, were chosen based on judgement. They can be set by editing the fftool source, namely the global variables BondTol and AngleTol. Use with care because spurious bonds and angles may be created if the tolerances are too large.

Improper Dihedrals

Improper dihedrals are often used to increase the rigidity of planar atoms (sp2) and differ from proper dihedrals in how they are defined. A proper dihedral i-j-k-l is defined between bonded atoms i-j, j-k, and k-l and corresponds to torsion around bond j-k, the dihedral being the angle between planes i-j-k and j-k-l. An improper dihedral i-j-k-l is defined between bonded atoms i-k, j-k and k-l, therefore k is a central atom bonded to the other three. The central atom of the improper dihedral is assumed to be the third in the list. Often in force fields the same potential energy function is used both for proper and improper torsions.

If improper records are supplied in a molecule file (in .zmat format) then those improper dihedrals are assumed by fftool. Otherwise, the script will search for candidate improper dihedrals on all atoms with three bonds, with any of .zmat, .mol, .pdb or .xyz input formats. A number of warning messages will be printed if there are atoms with three bonds, which can be ignored if the atoms in question are not centers of improper torsions. The number and order of the atoms in the true improper dihedrals should be checked in the files created.

Periodic Boundary Conditions

In molecular systems the initial configuration will generaly not contain molecules crossing boundaries of the simulation box. A buffer distance of 1.5 angstrom is reserved at the box boundaries to avoid overlap of molecules from periodic images in the initial configuration, as explained in the packmol documentation (this empty space is added by fftool only for orthogonal boxes). So the user should allow for this empty volume when supplying the size of the box.

For simulations with extended materials it is possible to create chemical bonds across boundaries. The option -p allows specification of periodic conditions along x, y, z or combinations thereof. It is important in this case to supply dimensions for the simulation box using the option -b <l> for a cubic box, or -b <lx,ly,lz> for a general orthogonal box, or -b <a,b,c,alpha,beta,gamma> for a general parallelepiped (triclinic box). An energy minimization step prior to the start of the MD simulation is highly recommended because no extra space is left near the boundaries and certain molecules may overlap with those of neighboring images.

The coordinates of the atoms of the material have to be supplied in .xyz format and prepared carefully so that distances across periodic boundaries are within the tolerance to identify bonds. The number of bonds in the output files created should be checked.

It is important that only the material for which bonds are to be established across boudaries is supplied in .xyz format. The initial files for other molecules in the system should be in .zmat or .mol formats, which contain connectivity information. This is to avoid spurious bonds between atoms of the molecular species that happen to be positioned too close to boundaries.

The pack.inp file will likely need manual editing in order to position the atoms of the material precisely.

Force Field File Format

The fftool script reads a database of molecular force field terms in the format described below. See the examples directory.

Blank lines and lines starting with # are ignored.

There are five sections, with headings ATOMS, BONDS, ANGLES, DIHEDRALS and IMPROPER. Under each section heading, registers concerning the different types of term in the force field are given.

ATOMS records describe, for each type of atom:

• the non-bonded atom type used for intermolecular interactions (these types may differ in the charges or intermolecular potential parameters)

• the bonded atom type used in intermolecular interactions (these types determine the intramolecular terms such as bonds, angles dihedrals)

• the mass in atomic units

• the electrostatic charge in elementary units

• the non-bonded potential type, e.g. lj

• potential parameters, namely Lennard-Jones sigma and epsilon

    C3H   CT  12.011  -0.18   lj    3.50   0.27614
  

BONDS records describe covalent bonds between intramolecular atom types:

• two bonded atom types

• type of bond potential, e.g. harm for harmonic potential or cons for a constrained bond.

• bond potential parameters, namely euqilibrium distance and force constant (the latter in the form k/2 (x - x0)^2)

    CT  CT   harm   1.529   2242.6
  

ANGLES records describe valence angles between intramolecular atom types:

• three bonded atom types, in which the central atom is bonded to the other two, e.g. i-j and j-k are bonded.

• type of angle potential, e.g. harm for harmonic potential or cons for a constrained angle.

• angle potential parameters, namely equilibrium angle and force constant (the latter in the form k/2 (x - x0)^2

    HC  CT  CT   harm   110.7   313.8
  

DIHEDRALS records describe torsion angles between intramolecular atom types:

• four bonded atom types, in which atoms i-j, j-k, k-l are bonded.

• type of dihedral potential, e.g. opls for OPLS cosine series with four terms.

• dihedral potential parameters, with the coefficients in the form V_n/2 (1 +/- cos(n phi)).

    CT  CT  CT  CT   opls    5.4392   -0.2092    0.8368    0.0000
  

IMPROPER records describe improper dihedral angles between intramolecular atom types:

• four bonded atom types, in which atoms i-k, j-k, k-l are bonded.

• type of dihedral potential, e.g. opls for OPLS cosine series with four terms.

• dihedral potential parameters, with the coefficients in the form V_n/2 (1 +/- cos(n phi)).

    CA  CA  CA  HA   opls    0.0000    9.2048    0.0000    0.0000
  

References

• Packmol: L. Martinez et al. J Comp Chem 30 (2009) 2157, DOI: 10.1002/jcc.21224

• LAMMPS: S. Plimton, J Comp Phys 117 (1995) 1, DOI: 10.1006/jcph.1995.1039

• DL_POLY: I.T. Todorov and W. Smith, Daresbury Lab.

• GROMACS: H.J.C. Berendsen, D. van der Spoel, R. van Drunen, Comp Phys Commun, 91 (1995) 43, DOI: 10.1016/0010-4655(95)00042-E