[gmx-users] Cannot minimiza - vdwradii.dat ???

[Justin Lemkul]

On 11/15/12 11:34 AM, Steven Neumann wrote:
> On Thu, Nov 15, 2012 at 4:03 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>>
>>
>> On 11/15/12 10:55 AM, Steven Neumann wrote:
>>>
>>> On Thu, Nov 15, 2012 at 3:47 PM, Justin Lemkul <jalemkul at vt.edu> wrote:
>>>>
>>>>
>>>>
>>>> On 11/15/12 10:41 AM, Steven Neumann wrote:
>>>>>
>>>>>
>>>>> atomtypes.atp
>>>>>
>>>>> FE      55.84700 ;      heme iron 56
>>>>>
>>>>> ffnonbonded
>>>>>
>>>>> FE     9      18.998    0.000   A     0.115816833358244  -0
>>>>>
>>>>> aminoacids.rtp
>>>>>
>>>>> [ FE ]
>>>>>     [ atoms ]
>>>>>           FE      FE      2.00    0
>>>>>
>>>>>
>>>>> Hmmmm, confusing...
>>>>>
>>>>
>>>> What force field is that?  None of those parameters seem to come from any
>>>> of
>>>> the default force fields in Gromacs. The entry in ffnonbonded is clearly
>>>> wrong, with the parameters appearing to belong to fluorine rather than
>>>> iron,
>>>> but the line appears to have been mangled.  It looks most like CHARMM,
>>>> but
>>>> someone has messed with ffnonbonded.itp.
>>>>
>>>> Regardless, if you look at the parameters for FE in any of the force
>>>> fields
>>>> in Gromacs, you will see that their nonbonded parameters (C6/C12 or
>>>> sigma/epsilon) are generally set to zero.  They interact only via ionic
>>>> interactions because they were likely only intended for use in cofactors
>>>> like heme, where other bonded parameters fix their geometry.  Using them
>>>> as
>>>> free ions makes little sense.
>>>>
>>>>
>>>> -Justin
>>>>
>>>
>>> This is charmm27 but modified as you can see. Shall I use:
>>>
>>> FE      26      55.84700        2.000   A       0.115816833358  0
>>> 0
>>>
>>> in original Gromacs charmm it is
>>>
>>> FE      26      55.84700        0.000   A       0.115816833358  0.0 ;
>>> partial charge def not found
>>>
>>
>> These changes will have no effect and will not solve your problem.  The
>> charge field in ffnonbonded.itp is never used for anything.  I don't know
>> what the extra zero is intended for, but it will probably break the ability
>> of grompp to read the file.
>>
>> Moreover, the point of my previous comments was that your use of this atom
>> type is not appropriate.  The FE atom type still has zero repulsive
>> interactions, unless explicitly calculated with each atom type.  As far as I
>> can tell from ffnonbonded.itp, this repulsive term is zero for every
>> interaction, meaning no matter what you do, your Fe ions will always collide
>> with something negatively charged because there are no LJ terms to impede
>> the Coulombic attraction, leading to an infinitely negative potential at a
>> fixed point and thus the crash.
>>
>> You need to find better parameters entirely if you wish to simulate free Fe
>> ions.
>>
>> -Justin
>
> Thank you for this. I will try to look at Charmm forum for those
> parameters. I was also thinking about parameters which Zn(2+) has got
> as they have similar mass. What do you think?
>

For almost every element in the periodic table, you can find another element 
with a similar mass.  That doesn't mean their nonbonded parameters will be of 
any use.  A 20% difference between iron and zinc is non-negligible, anyway ;) 
More importantly, transition elements are very challenging, and their behavior 
is governed by a lot more than just the naive MM representation of a sphere of 
fixed charge.

-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

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