[gmx-users] Re: What is the purpose of the [ pairs ] section?

[Justin Lemkul]

On 12/11/12 3:36 PM, Andrew DeYoung wrote:
> Hi Javier, Mark, Justin, and Bogdan,
>
> Thank you SO much for your time!  The responses that you gave are INCREDIBLY
> helpful to me.  I think that they will be very useful, too, to future users
> as helpful canonical answers.
>
> If you have time, I have another question related to the [ pairs ] section.
> I generated an .itp file for my large molecule (containing 492 atoms) using
> g_x2top, in the OPLS-AA force field.  g_x2top, it seems, specifies all 1-4
> interactions in the [ pairs ] section.
>
> The [ pairs ] section generated by g_x2top for my 492-atom molecule contains
> 2225 pairs.  I have not checked all 2225 of these pairs (and I should), but
> from looking at a subset of them, it appears that these all correctly
> correspond to 1-4 interactions, which as Bogdan pointed out, in the OPLS-AA
> would be the correct behavior.  (That is, in OPLS-AA, 1-4 interactions --
> and only 1-4 interactions -- are specified in [ pairs ], assuming that
> nrexcl = 3, which in my case it is.)
>
> Only the indices are specified in my [ pairs ] section, so I have set
> gen-pairs = yes to generate (calculate) the parameter combinations, as
> Bogdan explained.
>
> But what confuses me is the grompp output, which says:
>
> Generated 276 of the 276 non-bonded parameter combinations
> Generating 1-4 interactions: fudge = 0.5
> Generated 276 of the 276 1-4 parameter combinations
>
> Why is the number of nonbonded interactions 276, and to what does this
> refer?  In a previous post in this thread, Justin said that this 276 number
> refers only to 1-4 interactions.  If this is the case (and I DO completely
> trust Justin!), then why doesn't it say 2225 instead of 276?
>
> Maybe the 276 number refers to the number of UNIQUE combinations found
> within the 2225 pairs, but I don't see how this could be since my molecule
> only contains four atomtypes (two different carbon atomtypes, one oxygen
> atomtype, and one hydrogen atomtype).  Does anyone have any thoughts on
> this?
>

The numbers you get do indeed reflect only unique combinations, if my quick scan 
of the code is correct (grompp is an ugly beast).  It is important to remember 
that the total number of types is not simply those that you have used, but those 
belonging to the parent force field.  grompp will process all of that 
information.  So even if you're only using 4 types for the simulation, I suspect 
that you're actually using a modified version of OPLS-AA that has 23 atom types 
total, a subset of the whole force field.  The full OPLS-AA generates 332520 
combinations, and that's with 800+ atomtypes, even though my protein simulation 
(which reports 332520 combinations being generated) only uses a few dozen atomtypes.

-Justin

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

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

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