[gmx-users] question about preferred values in itp files

[Justin Lemkul]

On 4/13/16 1:53 PM, abhishek khetan wrote:
> Dear gmx-users,
>
> I have the opls-aa forcefield file for acetonitrile. It looks something
> like:
> -----------------------------------------------------------------------------------------------------------
>   [ moleculetype ]
>   ; Name            nrexcl
>   ACN                 3
>
>   [ atoms ]
>   ;   nr       type  resnr residue  atom   cgnr     charge       mass
>        1   opls_755      1    ACN      C1      1      -0.08    12.01100
>        2   opls_759      2    ACN      H2      1       0.06     1.00800
>        3   opls_759      3    ACN      H3      1       0.06     1.00800
>        4   opls_759      4    ACN      H4      1       0.06     1.00800
>        5   opls_754      5    ACN      C5      2       0.46    12.01100
>        6   opls_753      6    ACN      N6      3      -0.56    14.00670
> -----------------------------------------------------------------------------------------------------------
>
> My topol.top file looks like
> -----------------------------------------------------------------------------------------------------------
> #include "./oplsaa.ff/forcefield.itp"
> #include "./oplsaa.ff/acn.itp"
>
> [ system ]
> ; Name
> ACN216
>
> [ molecules ]
> ; Compound        #mols
> ACN              216
> -----------------------------------------------------------------------------------------------------------
>
> And the forcefield.itp looks like
> -----------------------------------------------------------------------------------------------------------
> [ defaults ]
> ; nbfunc    comb-rule    gen-pairs    fudgeLJ    fudgeQQ
> 1        3        yes        0.5    0.5
>
> #include "ffnonbonded.itp"
> #include "ffbonded.itp"
> -----------------------------------------------------------------------------------------------------------
>
> Needless to mention, that the ffbonded.itp and ffnonbonded.itp are the
> standard files from the oplsaa.ff directory in which the opls_XYZ
> [atomtypes]/[bondtypes]/[dihedraltypes] etc are defined.
>
> My questions now are these:
>
> 1. When the charge and mass are already mentioned in the ffbonded.itp and
> ffnonbonded.itp files for each of the opls_XYZ entries, then why do we
> specify it again in the acn.itp file? Which values are read in the
> simulations? the one from ffnonbonded.itp or the ones from acn.itp ? In
> acse the acn.itp values are preferred, does it mean that the epsilon and
> sigma value are still read from the ffnonbonded.itp file ?
>

Charges in ffnonbonded.itp are never used for anything.  Values specified in a 
topology always override those in ff(non)bonded.itp; this hierarchy is explained 
in the manual.

> 2. BIGGER QUESTION: When i look at the ffnonbonded and ffbonded files - the
> bondtypes are specified in terms of the atom symbols from atomtpes, instead
> of the . The ffnonbonded and non bonded have entries like:
> ------------------------------------------------------------------------------------------------------------
> [ atomtypes ]
>   opls_001   C    6      12.01100     0.500       A    3.75000e-01
> 4.39320e-01 ; SIG
>   opls_017   C    6      12.01100     0.700       A    3.75000e-01
> 4.39320e-01 ; SIG
>   opls_026   C    6      12.01100     0.265       A    3.75000e-01
> 4.60240e-01 ; SIG
> ..........
> [ bondtypes ]
>    C     C3      1    0.15220   265265.6   ; END
>    C     CA      1    0.14900   334720.0   ; wlj 8/97
>    C     CB      1    0.14190   374049.6   ; GUA
>    C     CM      1    0.14440   343088.0   ; THY
>    C     CS      1    0.14900   334720.0   ;
> ..........
> ------------------------------------------------------------------------------------------------------------
>
> As you see all the bondtypes involve C as one of the atom types - how does
> gromacs know which atomtype to choose ?? there are multiple atomtypes
> corresponding to the same symbol. Or are have the symbols been chosen in
> such a way that when wither if opls 001/017/026 bonds with C3 then the bond
> distance
>

The bonded parameter space is significantly smaller than the nonbonded parameter 
space.  The translations are in ffnonbonded.itp so grompp knows how to map 
bonded and nonbonded types.

> 3. EVEN BIGGER QUESTION: From what I understand, once I specify the
> opls_XYZ type for a certain atom in a given molecule along with
> [bonds]/[dihedrals]/etc in the acn.itp, it reads all specifications along
> with the equilibrium bond distances, the angles and all that stuff from the
> ffbonded.itp and ffnonbonded.itp files and constructs the equilibrium
> geometry of the molecule by itself. Does that mean that any geometry
> specified by me for a single molecule is henceforth irrelevant ? I have
> geometry for acetonitrile, calculated at a very high level of quantum
> chemistry, perhaps more accurate than the one supplied in the pdb file that
> came with the forcefield. However, this accurate geometry may not be
> exactly the same as that constructed by gromacs from the ffbonded.itp and
> ffnonbonded.itp. Why do then, people always provide a *pdb or *gro file for
> the molecule along with the itp. if one has the acn.itp and the
> ffbonded.itp and ffnonbonded.itp files, the geometry can already be
> constructed... or ?
>

Nothing in GROMACS constructs coordinates like this. Your input geometry is 
read. The energies and forces are calculated using the equilibrium parameters in 
the force field and your input coordinates.

-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

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