Dihedrals not found about foyer HOT 13 CLOSED

This isn't an issue with the Foyer code itself, rather that there are 6 dihedrals found in the molecule that haven't been found in the forcefield xml file. I.e., they need to be defined by haven't yet.

I'll note that in some cases you may have quartets that don't require dihedral parameters (especially for CHARMM as the dihedral parameters are corrections to the 1-4 interactions not a replacement). If that is the case, then you'd just want to modify your apply call to tell it to ignore this (i.e., raise a warning, but not throw an error):

typed_gly = charmm.apply(gly,verbose=True, assert_dihedral_params=False)

Bonds, angles, dihedrals, and impropers each have an associated assert flag that can be set (i.e., assert_bond_params, assert_angle_params, assert_dihedral_params, assert_improper_params). This is especially useful as, e.g., a CG forcefield might not need to define dihedral or angle parameters or one might use fictitious bonds in mbuild to orient building blocks in the system, but not want them to be actually chemically bound. See the documentation about this

I'm going to close this issue as it appears the code is functioning as expected, but send me a message if you have more questions.

from foyer.

@chrisiacovella I think there is more going on here than meets the eye. All of the missing dihedrals seem to be found in the forcefield. Additionally, the atomtyping works using the GMSO methodology:

from forcefield_utilities import FoyerFFs
loader = FoyerFFs()
gmso_ff = loader.load('ffs/charmm_foyer/glycine.xml').to_gmso_ff()
from gmso.parameterization import apply
from gmso.external.convert_mbuild import from_mbuild
gmso_top = from_mbuild(gly) #object created using above recipe
apply(gmso_top, gmso_ff, identify_connections=True)
print(gmso_top.is_typed)

Since this seems to work properly, there seems to be something more than just missing parameters causing the openmm issues seen here. I can spend a little bit of time better showing the dihedrals in the forcefield, and the identities of the missing dihedrals.

from foyer.

I checked this XML manually and the dihedral is actually there (for example: <Proper class1="O" class2="C" class3="CT2" class4="NH1" periodicity1="1" phase1="0.0" k1="0.0" />), but was just not being found. This might be on the foyer side (or our use of openmm/parmed).

from foyer.

glycine.txt

from foyer.

Heres the xml for this molecule

from foyer.

I added the line to the PeriodicTorsionForce and it still errors out with a missing dihedral

Proper class1="" class2="" class3="" class4="" periodicity1="1" phase1="0" k1="0"

from foyer.

Ah I was too quick in closing. So this is a ParmEd issue we have run into in the past. If the dihedral has all the phase parameters set to 0 (i.e., it doesn't modify anything), ParmEd ignores them. The check_dihedrals function does not ignore them and throws an error.

from foyer.

I added phase1="3.141592653589793" to all the ones with k1=0 and this error is the same as before, so I'm not sure this is the issue, provided this is what you meant

from foyer.

regardless of what you set phase1 to be if k1 is hero it doesn't modify it in anyway, so ParmEd ignores it.

from foyer.

OK, yeah it works if I change K=1 to K=0.0000000001 for ones with only 1 K-value

from foyer.

@chrisiacovella good call !!!

from foyer.

See the file for K=1 to K=0.0000000001 for ones with only 1 K-value

glycine_mod_ks_to_near_0_values.txt

from foyer.

This solution seems to work so I am closing this issue

from foyer.