2.3. MD Simulations with Alexandria Force Fields

2.3.1. Using ACT simulate

The command alexandria simulate will perform a simulation in the gas phase, i.e. without periodic boundary conditions. This obviously limits the usefulness but the OpenMM software can be used for condensed-phase simulations using ACT force fields [1, 25]. The alexandria simulate utility is aimed for validation of the potentials, for instance by evaluating whether the total energy is conserved, it can be verified that the force is the derivative of the potential.

The utility can also be used to minimize the energy of a molecule or small cluster

alexandria simulate -minimize -ff actff.xml -f coords.pdb -charges mp2.xml -c afterem.pdb -g minimize.log

where actff.xml is the force field file, coords.pdb is a structure file (xyz and sdf are supported as well). The alexandira -charges mp2.xml indicates a molprop file (database) with monomeric (optimized) structures used for generating charges. The files afterem.pdb and minimize.log are output files.

2.3.2. The ACT-OpenMM interface

The OpenMM [29] software in its native form is controlled from Python scripts. This allows users great control over the simulations. OpenMM allows to specify user-defined energy functions, which we use to implement both Gaussian-distributed charges [19, 30] and many Van der Waals potentials [25] (see Section Energy Function). To facilitate this, a special python code was implemented that makes it relatively easy for the user to run simulations and minimization. The first step is to convert an ACT force field file (actff.xml}) to one compatible with OpenMM (*openmmff.xml):

alexandria gentop -ff actff.xml -openmm openmmff.xml
-charges mp2.xml -db "water methanol"

with two additional arguments. First, flag -charges mp2.xml indicates a molprop file (database) with monomeric (optimized) structures used for generating charges and, second, the flag -db “water methanol” instruct the code to produce an OpenMM topology for those compounds. If one or more of the compounds is not present in the database, a warning will be issued.

2.3.2.1. MD simulations using OpenMM

It is easy to perform simulations based on an Alexandria force field. For this, we rely on the OpenMM software that you need to install separately. Then you can use the ACT python interface to OpenMM, as in this simple script, that we will call run.py:

#!/usr/bin/env python3

from act_openmm import ActOpenMMSim

sim = ActOpenMMSim(pdbfile="file.pdb",
                   datfile="dimer.dat",
                   xmlfile="ff.xml",
                   txtfile="output2.txt",
                   verbose=True)
sim.run()
sim.log_to_xvg("energy.xvg", [ "Potential Energy (kJ/mole)"])
sim.log_to_xvg("temperature.xvg", [ "Temperature (K)" ])
sim.log_to_xvg("density.xvg", [ "Density (g/mL)" ])

Here, we first instantiate an ActOpenMMSim object, and pass it a structure file file.pdb, a simulation parameter file dimer.dat and a force field file ff.xml. We also instruct the code that output should be written to output2.txt and that we want a lot of information in that file. Then we can run it, that’s all! The last three lines are for convenience of plotting the results. We can run this script using:

python ./run.py

or submitted to a cluster, preferably with GPUs available.

2.3.2.2. Energy Minimization using OpenMM

An energy minimization of the input structure can be done using this script:

#!/usr/bin/env python3

from act_openmm import ActOpenMMSim

sim = ActOpenMMSim(pdbfile="file.pdb",
                   datfile="dimer.dat",
                   xmlfile="ff.xml",
                   txtfile="output2.txt",
                   verbose=True)
sim.setup()
sim.minimize()
sim.write_coordinates("final.pdb")

which is run here using the same flags as the simulation. Note that details about simulations and minimization can be specified in the dimer.dat file.