[gmx-users] KCl forms crystals at c=0.15M with amber99sb-ildn

[Justin Lemkul]

On 5/11/18 11:19 AM, Mark Abraham wrote:
> Hi,
>
> On Fri, May 11, 2018 at 12:07 PM Daniel Bauer <bauer at cbs.tu-darmstadt.de>
> wrote:
>
>> Hi,
>>
>> thanks for the response! Now i know what I have to do :)
>>
>> So did I understand you right that the ion parameters that are currently
>> used in amber99sb by GROMACS are not identical with the ones implemented
>> in the current version of amber99sb used by AMBER?
>
> Please everybody be very careful with statements like this. Generally the
> definition of a force field should never change, so that results remain
> reproducible. This is why derivatives of forcefields have been made, while
> correcting some observed deficiencies. However, since e.g. the original
> validation of the AMBER ports within GROMACS were made against AMBER 8,
> there have been some corrections of typos, etc. made by AMBER developers,
> which AFAIK have all made their way to the GROMACS files.
>
>
>> If yes, whats the
>> reason for this (might be more of a question for Justin or Mark). Are
>> there any plans to change this in future releases?
>>
> First let's clarify whether anything really had its definition changed ;-)
> If there's a recognized variant, then e.g. someone might contribute it to
> http://www.gromacs.org/Downloads/User_contributions/Force_fields so that
> people using it understand that it's a case of "buyer beware." These days
> we feel it would not be appropriate to add support for a new force field in
> GROMACS (say, CHARMM36) unless that change came alongside scripts that
> would be able to verify that the implementation remains correct, even
> though both GROMACS and the software for which it was developed will have
> to change as time passes. (So that issues like
> https://redmine.gromacs.org/issues/2046 are found and prevented early.)

I would also add on that it is common to see AMBER force fields used 
with a wide range of different ion parameters, based on what individual 
authors/groups find to be most effective. There are many, diverging 
variants of AMBER biomolecular force fields as well as ion parameters. 
AMBER99SB formally refers to a variant of the protein force field, and 
is coupled with standard ions. This is the "default" AMBER approach 
unless updated ion parameters are then employed in conjunction with the 
biomolecular force field. It becomes somewhat hard for the GROMACS 
implementation to satisfy the flexibility that other proteins like AMBER 
and CHARMM offer in terms of user-selected parameters that can be chosen 
on the fly.

Perhaps the only thing we can change on our end is documentation, e.g. 
in forcefield.doc or the manual that says "standard ion parameters are 
used" and it's up to the user to employ something else that is better 
should s/he choose.

-Justin

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

Justin A. Lemkul, Ph.D.
Assistant Professor
Virginia Tech Department of Biochemistry

303 Engel Hall
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

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