[gmx-users] Exclude Coulomb interactions among certain atoms in SWM4 water molecule

[Justin Lemkul]

On 2/22/14, 4:10 AM, Sandipan Dutta wrote:
> Hi Justin,
>
> Thanks Justion for replying. I am currently simulating only 1 SWM4 water
> molecule and I am getting a few errors.
> I just want to clear up a few things below.
> _____________________________________________________________________
> *1.* when I run pdb2gmx with this .rtp file
>
> [ SM2 ]
>   [ atoms ]
>    OW1    WO    1.71636     0
>    HW2    WH    0.55733     0
>    HW3    WH    0.55733     0
>     DW    WD   -1.11466     0
>     SW    WS   -1.71636     0

You have an error here.  The Drude (DW) and lone pair (SW?) has incorrect 
charges.  The lone pair carries a -1.11466 charge; the Drude carries -1.71636.

>   [ bonds ]
>    OW1   HW2
>    OW1   HW3
>    OW1   DW
>    OW1   SW
>
> The .gro and .top file is created but I am getting this in the terminal
>
> Making cmap torsions...There are    0 dihedrals,    0 impropers,    6 angles
>               0 pairs,        4 bonds and     0 virtual sites
> Total mass 18.015 a.m.u.
> Total charge 0.000 e
>
> The DW and SW are virtual particles. But I am not sure why I am getting 0
> virtual sites. Though in the log file I see 1 shell and 1 Vsite.

A Drude particle is not a virtual particle; it has mass and charge but zero LJ, 
making it basically just another atom.  The SWM4-NDP lone pair is a virtual site.

> _____________________________________________________________________________
> *2.* I read in the manual that we don't to specify the angles for
> dummy/shell particles
> but if I keep the  OW1   DW and OW1   SW bond in .rtp file, when I run the
> grompp
> I get an error asking for the angles of these virtual particles. ( I found
> that the TIP4 virtual particle
> have some angles recorded in the [ angletypes ] section of ffbonded.itp )
>
> If I delete the OW1   DW and OW1   SW bonds in .rtp file
> then the output of pdb2gmx is
>
> Making cmap torsions...There are    0 dihedrals,    0 impropers,    1 angles
>               0 pairs,        2 bonds and     0 virtual sites
> Total mass 18.015 a.m.u.
> Total charge 0.000 e
>
> The non-hydrogen bonds are missing in the .top file and I get the error in
> the mdrun:
>
> Fatal error:
> Something weird with shells. They may not be bonded to something
>
> That is strange because even though I delete the .rtp bonds those bonds
> still appear in the ffbonded.itp file.
> So it looks like I can not delete the bonds of the virtual particles in
> .rtp file. Is that correct?
> __________________________________________________________________________
> *3.* Is this problem because of the forcefield? I am currently modifying
> gmx.ff.
>

Yes, almost certainly.  That force field is slated for removal because it is 
deprecated, and AFAIK is not related to SWM4 to gmx.ff in any way, so I doubt 
trying to hack it together under that force field will actually work.  It's 
fairly straightforward to implement it as part of the CHARMM Drude polarizable 
force field, which is actually what we are currently working on implementing in 
Gromacs.

-Justin

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

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

Department of Pharmaceutical Sciences
School of Pharmacy
Health Sciences Facility II, Room 601
University of Maryland, Baltimore
20 Penn St.
Baltimore, MD 21201

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

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