Currently, we have an issue where some converted AMBER protein force field files contain the prefixe protein- while others do not, causing seemingly straightforward operations like bringing in the phosphoaminoacid templates:

from simtk.openmm.app import ForceField
ff = ForceField('amber/protein.ff14SB.xml', 'amber/phosaa10.xml')

to fail:

Traceback (most recent call last):
 File “<stdin>“, line 1, in <module>
 File “/home/brucemah/miniconda3/envs/omm_source/lib/python3.6/site-packages/simtk/openmm/app/forcefield.py”, line 185, in __init__
  self.loadFile(files)
 File “/home/brucemah/miniconda3/envs/omm_source/lib/python3.6/site-packages/simtk/openmm/app/forcefield.py”, line 278, in loadFile
  template.atoms.append(ForceField._TemplateAtomData(atomName, typeName, self._atomTypes[typeName].element, params))
KeyError: ‘N’

We therefore need to decide whether we want to harmonize these by prefixing all protein atom types with protein- so these will be compatible.

We can support this by adding a line to the YAML:

  Prefix: protein

for each file we want to convert and prepend prefixes for atom types.

Questions:

• Which files should we do this with?
• Should we also do this with all files in oldff/?
• How should we augment the tests to catch these incompatibilities

cc: @rafwiewiora @hannahbrucemacdonald @peastman

We should likely include the fixed parameters and information on how to use the small molecule parameters in an upcoming OpenMM release.