[gmx-users] changing atom types versus adding dihedrals to atom types

[Justin Lemkul]

On 9/12/13 7:04 PM, Rafael I. Silverman y de la Vega wrote:
> I see,
>   when you say thorough parametrization and validation, what do you mean?
> How does one validate the system, calculate free energy in solution and
> compare it to tabulated data?
>

The target data depend on the force field for which parameters are being 
developed.  Matching QM geometries, water interaction energies and distances, 
free energies of solvation, crystal structures, liquid properties, etc. are all 
useful depending on the underlying force field and how it was derived.  For 
bonded parameters, matching QM energy scans and/or vibrational properties are 
common.

-Justin

>
> On Wed, Sep 11, 2013 at 4:57 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>
>>
>>
>> On 9/11/13 6:07 PM, Rafael I. Silverman y de la Vega wrote:
>>
>>> Hi all,
>>> in another thread it was recommended that instead of changing the atom
>>> types by analogy to ones with the desired parameter already assigned, that
>>> instead one should assign bonded parameters by analogy.  I was just
>>> wondering how this is better?
>>>
>>
>> I didn't suggest that.  I said that any assignment by analogy can have
>> negative outcomes.  Bonds and angles, in theory, are a bit easier to deal
>> with because the nonbonded interactions between their atoms are excluded.
>>   Torsions are very difficult because the torsional bonded terms have to be
>> balanced against the charge and LJ parameters of atoms separated by 3
>> bonds, i.e. 1-4 interactions.
>>
>> Let me state very clearly that, for any missing parameters, anything short
>> of a thorough parametrization and validation is (in my purist mind) unsound.
>>
>>
>>     As long as the atom type isnt too different, then why would the torsions
>>> not work out right? If the parameters are interrelated, what makes it
>>> better to add one term to an atom type rather than changing all the terms
>>> (to ones that are supposed to be together)?
>>>
>>>
>> The key is being sufficiently similar, but even subtle differences between
>> LJ parameters have very large consequences, especially for 1-4
>> interactions, so I find it hard to believe that one will recapitulate
>> accurate interactions in this case.  Of course, all assignments by analogy
>> bear some penalty for nonequivalent groups.  I balked at your initial post
>> in the other thread because there was no need to mess with atom types.  The
>> system at hand there contained protein and lipid and could be completely
>> described by existing atom types and bonded types, the OP there was just
>> (and perhaps still is) constructing the force field incorrectly.  I felt it
>> need to be stated that the approach there would not necessarily be
>> productive, and to make sure the archive was complete so that no one would
>> come across a post in vacuum and assume that he or she can simply swap
>> around atom types to make error messages go away.  Your initial statement
>> referred only to bonds and angles, which indeed would be insensitive to
>> changes in LJ parameters on any of those atoms, but since the thread was in
>> the context of dihedrals, I thought I should state my opinion there.
>>
>> -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 <jalemkul at outerbanks.umaryland.edu> | (410)
>> 706-7441
>>
>> ==============================**====================
>> --
>> gmx-users mailing list    gmx-users at gromacs.org
>> http://lists.gromacs.org/**mailman/listinfo/gmx-users<http://lists.gromacs.org/mailman/listinfo/gmx-users>
>> * Please search the archive at http://www.gromacs.org/**
>> Support/Mailing_Lists/Search<http://www.gromacs.org/Support/Mailing_Lists/Search>before posting!
>> * Please don't post (un)subscribe requests to the list. Use the www
>> interface or send it to gmx-users-request at gromacs.org.
>> * Can't post? Read http://www.gromacs.org/**Support/Mailing_Lists<http://www.gromacs.org/Support/Mailing_Lists>
>>

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

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

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