[gmx-users] Atom types comparison between CGenFF and Charm36FF

[Justin Lemkul]

On 6/27/17 3:27 PM, Mohsen Ramezanpour wrote:
> Dear Gromacs Users,
> 
> I am trying to parameterize a molecule in Charmm36FF.
> 
> As part of this molecule, there is a "neutral trimethylamine nitrogen" and
> its protonated case. i.e. C-N-C(2) and C-N(H)-C(2), respectively.
> 
> To have some idea about the appropriate atom types, I first used the
> GAAMP server.
> 
> Here are the assigned atom types by GAAMP (based on CGenFF force field):
> 
> 
> *For Neutral case:*
> NG301 for N,
> CG3AM0 for two methyl group connected to the N, i.e for bolded Carbons in
> C-N-*C(2).*
> HGA3 for each Hydrogen atom in CG3AM0 methyl groups,
> CG321 for the carbon atom connecting the NC(2) to the rest of molecule.
> i.e. for bolded Carbon in *C*-N-C(2),
> and HGA2 for the Hydrogen atoms connected to CG321.
> ###############################
> 
> *For Protonated case:*
> NG3P1 for N,
> CG334 for two methyl group connected to the N, i.e for bolded Carbons in
> C-N(H)-*C(2).*
> HGA3 for each Hydrogen atom in CG334 methyl groups,
> CG324 for the carbon atom connecting the NC(2) to the rest of molecule.
> i.e. for bolded Carbon in *C*-N(H)-C(2),
> and HGA2 for the Hydrogen atoms connected to CG321.
> ################################
> 
> Now, I want to find the best equivalent atom type in Charmm36FF for each
> case.
> 
> For the protonated case, I think it is easier as it is something between PE
> and PC lipid headgroups. For the neutral case, however, it is difficult to
> find similar atom types in Charmm36.
> 
> My approach was to check the LJ and bonded parameters for the assigned atom
> types (which are in CGenFF) and find the atom types in Charmm36 with the
> exact same values. Although possible for some cases, but there are many
> problems with this approach:
> 
> 1) the exact values are not found in Charmm36FF.
> 2) If it is found, there are a lot of parameters missing in the Charmm36FF
> force field. Not all the bonds, angles, and dihedrals are defined in
> Charmm36FF.
> 
> Thanks in advance for any comment or suggestion.
> 

The "G" in CGenFF is for "general," which means the parameters are not 
necessarily the same as the parent CHARMM force field, and are subsequently a 
compromise between being highly optimized (e.g. CHARMM36) and being broadly 
applicable (general) such that the types can be used across different molecules.

If you want to parametrize something for CHARMM, e.g. to be merged into a larger 
molecule, you're wasting your time with trying to generate a CGenFF topology and 
try to find exact matches in CHARMM36.  By definition, you won't.  If your goal 
is a CHARMM topology, then start from existing CHARMM atom types, import charges 
by analogy, and do a full parametrization procedure (well described in numerous 
places in the literature).

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Ruth L. Kirschstein NRSA Postdoctoral Fellow

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 629
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

jalemkul at outerbanks.umaryland.edu | (410) 706-7441
http://mackerell.umaryland.edu/~jalemkul

==================================================