Support for OpenMM

[orionarcher]

Summary

Include a summary of major changes in bullet points:

• setup functions for OpenMM simulations
• functions to anneal and equilibrate pressure in OpenMM simulations
• includes a fix to packmol also present in PR PackmolBoxGen incorrectly adding " to input script #3

Additional dependencies introduced (if any)

openmm: this is code for OpenMM IO

openff: needed to parameterize Sage Force Field

parmed: used to create suitably sized PDB files. Later may be used to add more force fields.

openbabel + rdkit are used, but they are already used in pymatgen.

TODO (if any)

• better support for water models, at least allow the user to specify TIP3 or TIP4.

• automatic partial charge assignment. It should be possible to include RESP partial charges in the parameterization.

• allow partial charge scaling as a setting in the generator

Checklist

• Code is in the standard Python style. The easiest way to handle this
  is to run the following in the correct sequence on your local machine. Start with running
  black on your new code. This will automatically reformat
  your code to PEP8 conventions and removes most issues. Then run
  pycodestyle, followed by
  flake8.
• Docstrings have been added in the Google docstring format.
  Run pydocstyle on your code.
• Type annotations are highly encouraged. Run mypy to type check your code.
• Tests have been added for any new functionality or bug fixes.
• All linting and tests pass.

[rkingsbury]

Thanks @orioncohen ! I won't be able to look at this in much detail until early next week but I'm excited to see how your OpenMM support is shaping up.

[orionarcher]

Sounds good!

I believe that the OpenMM IO can fit into the new IO abstraction pretty cleanly by serializing all of the OpenMM objects needed to create a simulation

The InputSetGenerator will be instantiated by user defined settings and create a Topology, System, Integrator, and Simulation. It will then serialize all of those objects into XML and store them. When get_input_set is called, it will create an InputSet that has the XML files as attributes. This process matches the rest of Pymatgen IO and is fit for HTMD.

The InputSet can then have a get_simulation method that instantiates a simulation and returns it. This way, no OpenMM objects are ever stored and the InputSetGenerator and InputSet remain MSONable.

[orionarcher]

What should determine if a parameter goes in InputGenerator.__init__ or InputGenerator.get_input_set?

The documentation seems to indicate that only the structure and other critical parameters should go in get_input_set, but I can imagine many situations where that would be inconvenient. For example, if I wanted to repeat a simulation at four different temperatures. I'm tempted to put all simulations settings in both __init__ and get_input_set for maximum flexibility, but I'm curious to hear more about the intended design.