OpenMM AGBNP plugin

A plugin that implements the AGBNP1 implicit solvent model [1] for OpenMM.

The plugin supports the OpenCL and Reference platforms.

This implementation continues the support for the GaussVol model [3], previously maintained here.

Implementation of the AGBNP2 model [2] is in progress.

Emilio Gallicchio [email protected]

Last Modified: November 2018

License

This software is released under the LGPL license. See LICENSE.

Credits

This software is written and maintained by Emilio Gallicchio [email protected].

Support from the National Science Foundation (ACI 1440665) is acknowledged.

The plugin interface is based on the openmmexampleplugin by Peter Eastman.

Rquirements

Requires OpenMM 7.2.2 or later. Last tested with OpenMM 7.2.2.

Installation

These instructions assume Linux. Install OpenMM 7; the easiest is through miniconda using these instructions. Install swig through conda as well:

conda install -c omnia openmm swig

Locate the OpenMM installation directory, otherwise it will default to /usr/local/openmm. If OpenMM was installed via conda the OpenMM installation directory will be something like $HOME/miniconda2/pkgs/openmm-7.2.2-py27_0

Download this plugin package from github:

git clone https://github.com/egallicc/openmm_agbnp_plugin.git

Build and install the plugin with cmake. For example, assuming a unix system and a conda environment:

. ~/miniconda2/bin/activate
mkdir build_openmm_agbnp_plugin
cd build_openmm_agbnp_plugin
ccmake -i ../openmm_agbnp_plugin

Hit c (configure) until all variables are correctly set, then g to generate the makefiles. OPENMM_DIR should point to the OpenMM installation directory. CMAKE_INSTALL_PREFIX normally is the same as OPENMM_DIR. The plugin requires the python API. You need python and swig to install it.

Once the configuration is done do:

make
make install
make PythonInstall

The last two steps may need superuser access depending on the installation target, or use the recommended conda environment.

Test

cd to the directory where you cloned the openmm_agbnp_plugin sources. Then:

cd example
export OPENMM_PLUGIN_DIR=<openmm_dir>/lib/plugins
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:<openmm_dir>/lib:<openmm_dir>lib/plugins
python test_agbnp.py

where <openmm_dir> is the OpenMM installation directory.

C++ API

#include "AGBNPForce.h"
AGBNPForce* force = new AGBNPForce();
force->setNonbondedMethod(CutoffNonPeriodic);//NoCutoff also accepted
force->setCutoffDistance(1.2);
force->setVersion(1); //set version to 0 for GaussVol
system.addForce(force);
for(int i=0;i<numParticles;i++){
   force->addParticle(radius[i], gamma[i], alpha[i], charge[i], ishydrogen[i]);      
}

• radius: van der Waals atomic radius (double)
• gamma: surface tension parameter (double)
• alpha: solute-solvent dispersion interaction parameter (double)
• charge: atomic charge in atomic units (double)
• ishydrogen: whether the atom is a hydrogen atom (bool)

Units: kJ/mol and nanometer.

Python API

from AGBNPplugin import AGBNPForce
gb = AGBNPForce()
gb.setNonbondedMethod(CutoffNonPeriodic) #NoCutoff also accepted
gb.setCutoffDistance(1.2 * nanometer)
gb.setVersion(1) #set version to 0 for GaussVol
for i in range(numParticles):
   #only the atomic radius is relevant for GaussVol 
   gb.addParticle(radius[i], gamma[i], alpha[i], charge[i], ishydrogen[i])
sys.addForce(gb)

The meaning of the parameters is the same as for the C++ API above.

Relevant references:

1. Gallicchio E., and R.M. Levy. AGBNP, an analytic implicit solvent model suitable for molecular dynamics simulations and high-resolution modeling, J. Comp. Chem. 25, 479-499 (2004).
2. Gallicchio, Emilio, Kristina Paris, and Ronald M. Levy. "The AGBNP2 implicit solvation model." Journal of chemical theory and computation 5.9 (2009): 2544-2564.
3. Baofeng Zhang, Denise Kilburg, Peter Eastman, Vijay S. Pande, Emilio Gallicchio. Efficient Gaussian Density Formulation of Volume and Surface Areas of Macromolecules on Graphical Processing Units. J. Comp. Chem., 38, 740-752 (2017).