[gmx-users] Regarding SwissParam topology

[Justin Lemkul]

On 6/10/15 1:36 AM, soumadwip ghosh wrote:
> Hi,
>     I have a question about the topology made by SwissParam which is
> compatible with CHARMM 27 force field. I hate to bring the same thing up
> again and again but I am really doubtful about some of the parameters
> created by the above mentioned web server. I have built a topology for
> fullerene molecule using it at it looks like
>
> [ moleculetype ]
> ; Name nrexcl
>   LIG      3
>
> [ atoms ]
> ; nr type resnr resid atom cgnr charge mass
>     1 CB   1  LIG C       1  0.0000  12.0110
>     2 CB   1  LIG C1      2  0.0000  12.0110
>     3 CB   1  LIG C2      3  0.0000  12.0110
>     4 CB   1  LIG C3      4  0.0000  12.0110
>     5 CB   1  LIG C4      5  0.0000  12.0110
>     6 CB   1  LIG C5      6  0.0000  12.0110
>     7 CB   1  LIG C6      7  0.0000  12.0110
>     8 CB   1  LIG C7      8  0.0000  12.0110
>     9 CB   1  LIG C8      9  0.0000  12.0110
>    10 CB   1  LIG C9     10  0.0000  12.0110
>    11 CB   1  LIG C10    11  0.0000  12.0110
>    12 CB   1  LIG C11    12  0.0000  12.0110
>    13 CB   1  LIG C12    13  0.0000  12.0110
>    14 CB   1  LIG C13    14  0.0000  12.0110
>    15 CB   1  LIG C14    15  0.0000  12.0110
>    16 CB   1  LIG C15    16  0.0000  12.0110
>    17 CB   1  LIG C16    17  0.0000  12.0110
>    18 CB   1  LIG C17    18  0.0000  12.0110
>    19 CB   1  LIG C18    19  0.0000  12.0110
>    20 CB   1  LIG C19    20  0.0000  12.0110
>    21 CB   1  LIG C20    21  0.0000  12.0110
>    22 CB   1  LIG C21    22  0.0000  12.0110
>    23 CB   1  LIG C22    23  0.0000  12.0110
>    24 CB   1  LIG C23    24  0.0000  12.0110
>    25 CB   1  LIG C24    25  0.0000  12.0110
>    26 CB   1  LIG C25    26  0.0000  12.0110
>    27 CB   1  LIG C26    27  0.0000  12.0110
> The parameters are properly placed in the latter part (kb, force
> constant, fu th0 kth ub0 kub th0 kth ub0 kub and so on) in appropriate
> directives [ angles ], [ dihedrals ] etc. But the CB atomtype does not
> exist as such in the CHARMM 27 ff directory. So how will the atomic

It's a new atom type defined in the SwissParam topology.  If it were missing, 
grompp would fail with a fatal error.  The "CB" type should probably be similar 
(if not identical) to the existing CA type, which is an aromatic carbon.

> interactions are validated in this case. I am stuck with this problem for a
> long time since I generally
> 1. draw the structure in gaussview 09
> 2. save it as mol2 file
> 3. Upload it on SwissParam web server to obtain topology.
>
> Is there anything wrong happening or is there some trivial point I'm
> missing out here? At the end of the day, my simulations do run without any
> hitch but I struggle to get the desired molecular interaction most of the
> time. Is it because the atomtype description itself is vague here and does
> not belong to CHARMM 27 ?
>

It's a new atom type provided by the server, but you'll have to read whatever 
the associated literature is to understand how atom typing is done.  Many such 
algorithms create "new" atom types that are actually just clones of existing 
types, but they are written in a topology so that it is self-contained and does 
not require the user to provide any other outside parameters.  It's for convenience.

> If this is indeed the problem, I am thinking of switching over to ParamChem
> in order to access CGenFF parameters.
>

Worth a shot, but for something like a fullerene, there's little to be gained 
here.  You will always have uncharged, aromatic types.

Take a step back and work with something you know the answer to so you can 
verify your method.  You need a defined system that has a known binding 
affinity.  If you can reproduce that, your method is reasonable.  If you're then 
getting weird stuff happening with untested systems, you'll at least have an 
idea of whether or not it's the right answer (even if it's unexpected) or you 
need further force field refinement.  ParamChem is what I always endorse because 
it provides penalty scores for all the parameters to alert you to possible 
deficiencies.  I don't see much gain trying to run a fullerene through 
ParamChem; the topology is trivially simple (it's all the same atom type and 
there are no charges) so you're not likely to get a much better description of 
it.  Much of what you're seeing could be attributed to a crude interpretation of 
what the fullerene parameters are.

-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

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