[gmx-users] GROMOS53a6 c12 of CH2 and CH3 in pairtypes incosistent with FF paper

[Justin Lemkul]

On 9/5/13 12:53 AM, Dallas Warren wrote:
> Going through the GROMOS53a6 parameters, found that there appears to be an inconsistency between what is present within the ffnonbonded.itp file and that quoted within the paper (Ooostenbrink et al 2004 http://dx.doi.org/10.1002/Jcc.20090) for the c12 LJ values for CH2 and CH3 in the [ pairtypes ] section.
>
> [ pairtypes ]
> ; i    j func          c6           c12
> ....
>        OA       OA  1  0.0022619536  1.265625e-06
> ....
>       CH2      CH2  1  0.0047238129  4.7419261e-06
> ....
>       CH3      CH3  1  0.0068525284  6.0308652e-06
>
> Then taking these c12 parameters from the ffnonbonded.itp file and converting to square root values so can compare with the values presented in the FF paper, Table 9.  Have also done the reverse, converting the Table 9 c12 values to the squared value so can compare with the value within the ffnonbonded.itp
>
> OA - OA
> ffnonbonded.itp	1.265625e-06  =>	0.001125000
> Table 9		1.265625E-06  <=  0.001125
>
> CH2 - OA
> ffnonbonded.itp	4.7419261e-06 =>	0.002177596
> Table 9		4.743684E-06  <=  0.002178
>
> CH3 - OA
> ffnonbonded.itp	6.0308652e-06 =>	0.002455782
> Table 9		6.031936E-06  <=	0.002456
>
> As you can see, the parameters for OA are consistent, while those for CH2 and CH3 are not.
>
> So, I suppose the questions are:
>
> 1) where did the values presented in the ffnonbonded.itp actually come from?
> 2) why are they not consistent with those that are defined in the FF paper, considering that the other cases are i.e. OA?
> 3) how much of a concern it that the values are that different (4th significant figure)?
>

The values in ffnonbonded.itp came from Gromos96 force field files via a script 
that did the unit conversion.  If you go to the ATB site, you can find the 
original 53a6 force field files and you will see that the values in 
ffnonbonded.itp are actually correct.  There are more significant figures for 
some of the terms than are presented in the paper.  Both of the values you have 
shown above, back-calculated from ffnonbonded.itp, actually reproduce the C12 
values in these files exactly.

-Justin

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

Justin A. Lemkul, Ph.D.
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

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