[gmx-users] Setting custom atoms in FF

[Justin Lemkul]

On 3/29/15 5:01 PM, Alex wrote:
> Whooooa, that's a huge blow, and it explains a bunch of other issues. I
> was thinking the difference had to be around 2-3% and set my pdb files
> very precisely.
>
> I actually think that either the tolerance is even larger, or there's something on top of that in the code.
>

The tolerance is 10%, used uniformly throughout the code (pdb2gmx.c, specbond.c, 
and x2top.c).  Though once an atom is assigned its status in g_x2top, it doesn't 
get overridden (IIRC) so your entries need to be unique and order of appearance 
can matter.

> Setting
>
> CJ   opls_996    0      12.011  3    C  0.10  C 0.10  C 0.10   ;bulk C
>
> still assigns #996 to bulk graphene carbons, despite the
> difference of about 42%. Anyway, this is nearly catastrophic for what I'm trying to do.
>  From what I understand, the tolerance cannot be set in the x2top
> command itself... Oy.
>

g_x2top also doesn't purport to be terribly smart.  It may be better to find 
another means to generate your topology if there are going to be complex systems 
and subtle differences.

> If you know it off the top of your head, could you please point me to
> the variable name in g_x2top.c (if it's there)? I'm using a
> precompiled version right now, but I guess I will have to mess with the code on our
> cluster.

Function is_bond() in x2top.c

>
> Allowing this tolerance to be set either as percentage, or in
> distance units would be extremely useful. One could always use the
> default value if no setting is entered.
>
> Could the last point be relayed to the developers? Sounds like a huge
> necessity to me, to be honest.
>

All of the developers are on this list and feature requests can be filed at 
redmine.gromacs.org.  Tunable tolerance would be interesting and probably not 
hard to implement, but it wouldn't necessarily just affect g_x2top.  A 10% 
tolerance is very permissive for anything run through pdb2gmx, for instance.

-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

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