Is OpenFF 2.0 compatible with AMBER ff19sb? and if yes how to combine the protein and ligand prmtop file?

[aniketsh]

Is OpenFF 2.0 compatible with AMBER ff19sb? and if yes how to combine the protein and ligand prmtop file?

[j-wags]

Hi @aniketsh,

Is OpenFF 2.0 compatible with AMBER ff19sb?

I believe that they are compatible, since both ff19sb and openff-2.0.0 were fit using similar methods to a mix of QM and physical property data, and both force fields use compatible functional forms.

if yes how to combine the protein and ligand prmtop file?

This is the hard part - My understanding is that we generally recommend use of ff14sb or earlier, since ff19sb introduces "CMAP" terms which aren't consistently supported in other force field definitions/simulation engines. I think that OpenMM may be able to handle them, but don't have first-hand experience. So sander/pmemd may be the best simulation engines of choice to ensure that the terms are interpreted correctly.

Generally, there are two options for making systems that use a mix of multiple force fields:

1. Use openmmforcefields to assign parameters to everything, then export the resulting OpenMM System to a ParmEd Structure and convert that to your format of choice
2. Assemble the different components using the tool that's best suited for the job, and then combine them using ParmEd.

Since openmmforcefields doesn't seem to support ff19sb yet, I think that approach 2 is the only option right now. It would look something like this example

• parameterize the protein using tleap, writing the result to protein.prmtop and protein.inpcrd files
• parameterize the ligand using OpenFF, getting an OpenMM System object as a result (you can either do the solvent and ions using tleap or the OpenFF toolkit, though note that openff-2.0.0 is trained to be compatible with TIP3P water. The toolkit will apply TIP3P parameters for water as part of the openff-2.0.0 force field, but if you have tleap handle the water be sure to specify TIP3P)
• turn the OpenMM System into a ParmEd Structure
• loading protein.prmtop and protein.inpcrd into another ParmEd Structure
• Use ParmEd's + operator to make the complex: complex_structure = protein_structure + ligand_structure

I'd note, however, that you should look very closely at the results of this conversion. All of the CMAP-handling functionality in OpenMM and ParmEd seems to be just a few months old, so there may be conversion/interpretation bugs that people haven't found yet!

[aniketsh]

Thanks Jeff. I use parmed usually to combine the system and it had worked before. With CMAP though it works in only one direction.

Parmed version (3.4.3)

import parmed as pmd

ptn=pmd.amber.AmberParm(prm_name='host.prmtop',rst7_name='host.crd')
lig=pmd.amber.AmberParm(prm_name='MOL.prmtop',rst7_name='MOL.inpcrd')

system=ptn+lig # This works
system=lig+ptn # However this doesn't

system.save('sys.prmtop',overwrite=True)
system.save('sys.crd',overwrite=True)

[mattwthompson]

system=ptn+lig # This works
system=lig+ptn # However this doesn't

This looks like a ParmEd bug of some sort, could you open an issue there?