Re: [AMBER] igb=7, 8 incompatible with the ILDN modifications of ff99SB

From: Vlad Cojocaru <vlad.cojocaru.mpi-muenster.mpg.de>
Date: Tue, 11 Sep 2012 16:15:29 +0200

Jason,

As always, a great, detailed answer from you which helps a lot. Thanks a
lot. I actually started looking at this based on a previous discussion
we had about MMPBSA where we also had a problem with these atom types

Best,
Vlad


On 09/11/2012 03:50 PM, Jason Swails wrote:
> The choice of the C3 atom type for ildn was unfortunate, since I think some
> of the (old?) Amber force fields had previous plans with it. The code to
> assign radii is found in 2 places: LEaP and ParmEd. The code in ParmEd is
> a bit more straightforward, since it is separated out by radius set, and
> LEaP converts the bondi set to an internal code which is not entirely
> obvious.
>
> The radii are assigned in leap, but beginning with AmberTools 12 they can
> actually be changed via ParmEd. The problem in your case is that the ILDN
> modifications changed CG2 in ILE from CT (radius 1.8) to C3 (radius 2.2).
> (The Python version of the radius assignment code is in
> $AMBERHOME/AmberTools/src/parmed/ParmedTools/changeradii.py).
>
> Probably the easiest thing to do (since this radius was intended to be 1.8)
> is to physically change the CG2 radius to 1.8 in the topology file. This
> is very easy with ParmEd (either the GUI xparmed.py or the CLI parmed.py).
>
> Look for the "change" command, then change the RADII of :ILE.CG2 to 1.8
> (then, of course, write out a new topology file). With parmed.py, it would
> look something like this:
>
> parmed.py your.prmtop << EOF
> changeRadii mbondi3
> change RADII :ILE.CG2 1.8
> parmout new.prmtop
> EOF
>
> The first command makes sure that you start from the mbondi3 radius set
> (unnecessary if you set it in leap). The second command, "change", will
> allow you to change any atomic properties (e.g., ATOM_NAME, CHARGE, MASS,
> RADII, etc.), so we change the RADII property of a given mask to the
> assigned value. This will change every ILE.CG2 atom in your system to 1.8.
>
> HTH,
> Jason
>
> On Tue, Sep 11, 2012 at 9:14 AM, Vlad Cojocaru <
> vlad.cojocaru.mpi-muenster.mpg.de> wrote:
>
>> The same problem occurs with igb=8 and mbondi3, the CG2 atom in ILE (C3
>> type) is assigned a radius greater than 2 and sander exits with same
>> error I reported for igb=7 and the "bondi" radii set.
>
>> Of course checking the topology file for all assignments is rather
>> tedious and I would be happy to avoid that if possible. So, probably the
>> most easy solution becomes to use the original ff99SB force field for
>> the implicit simulations and switch to the desired modified force field
>> for the subsequent explicit solvent simulations. Not a very elegant
>> solution but I guess an acceptable one for now.
>>
>> Can anybody tell me where under the amber source tree are these radii
>> sets stored ? Are these hard coded ? I could not find any file that
>> contains the radii sets ..
>>
>> Best
>> Vlad
>>
>> On 09/10/2012 03:25 PM, Carlos Simmerling wrote:
>>> oh yeah, I see it now. that's what I get for trying to answer questions
>>> early in the AM!
>>> I do not at all recommend igb=7. the underlying model is good, but the
>>> parameters were not well optimized. We refit them for igb=8 and the
>>> performance is far better. you'll also want mbondi3 radii.
>>>
>>> the easiest way for you to debug this is to actually look at the radii
>> that
>>> leap assigned in each of the cases. It could indeed be that the ILDN atom
>>> types aren't compatible with the radii logic in leap- typically leap does
>>> need to know more than just the element in order to set the radii.
>> getting
>>> an error is the best case- it's possible to set them incorrectly for your
>>> molecule and actually get bad behavior without an error.... leap really
>>> doesn't know that much about chemistry.
>>>
>>> On Mon, Sep 10, 2012 at 9:20 AM, Vlad Cojocaru <
>>> vlad.cojocaru.mpi-muenster.mpg.de> wrote:
>>>
>>>> Carlos,
>>>>
>>>> I mentioned already I am using AMBER 12 ... But it was kind of between
>>>> the lines .. Sorry for that !
>>>>
>>>> Best,
>>>>
>>> Vlad
>>>> On 09/10/2012 02:57 PM, Carlos Simmerling wrote:
>>>>> how radii are set does depend on which version of Amber you have, so
>> we'd
>>>>> need to know that before we can help...
>>>>>
>>>>> On Mon, Sep 10, 2012 at 4:49 AM, Vlad Cojocaru <
>>>>> vlad.cojocaru.mpi-muenster.mpg.de> wrote:
>>>>>
>>>>>> Dear all,
>>>>>>
>>>>>> It appears that the Generalized Born simulations with igb=7 or igb=8
>> are
>>>>>> incompatible with the ILDN modification of the ff99SB force field.
>>>>>> I was trying to do an implicit solvent optimization of an unstructured
>>>>>> tail before starting explicit solvent simulations, but after setting
>>>>>> PBradii to "bondi" in Leap and running with igb=7 (AMBER 12), I got
>> the
>>>>>> error
>>>>>>
>>>>>> " Atom 25 has radius 2.2000000000000002 outside of
>>>>>> allowed range
>>>>>> of 1.0 to 2.0 angstroms for igb=7. Regenerate prmtop file with
>> bondi
>>>>>> radii."
>>>>>>
>>>>>> Looking in the topology I realized that this is a CG2 atom in ILE
>> which
>>>>>> has the type C3 ...
>>>>>>
>>>>>> If I do the same simulation with igb=5, I do not get any error .. But
>>>>>> now I am actually wondering if the radii for these atom types are
>>>>>> assigned properly with the set "mbondi2" (which I used for the igb=5
>>>>>> simulations)...
>>>>>>
>>>>>> I know that ILDN modifications are optimized for explicit solvent
>>>>>> simulations, but since my goal is actually explicit solvent
>> simulations
>>>>>> I did not want to use a different force field for the intermediate
>> step
>>>>>> of implicit solvent optimization.
>>>>>>
>>>>>> Best wishes
>>>>>> Vlad
>>>>>>
>>>>>> --
>>>>>> Dr. Vlad Cojocaru
>>>>>> Max Planck Institute for Molecular Biomedicine
>>>>>> Department of Cell and Developmental Biology
>>>>>> Röntgenstrasse 20, 48149 Münster, Germany
>>>>>> Tel: +49-251-70365-324; Fax: +49-251-70365-399
>>>>>> Email: vlad.cojocaru[at]mpi-muenster.mpg.de
>>>>>> http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
>>>>>>
>>>>>>
>>>>>> _______________________________________________
>>>>>> AMBER mailing list
>>>>>> AMBER.ambermd.org
>>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>>
>>>>> _______________________________________________
>>>>> AMBER mailing list
>>>>> AMBER.ambermd.org
>>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>>
>>>>>
>>>> --
>>>> Dr. Vlad Cojocaru
>>>> Max Planck Institute for Molecular Biomedicine
>>>> Department of Cell and Developmental Biology
>>>> Röntgenstrasse 20, 48149 Münster, Germany
>>>> Tel: +49-251-70365-324; Fax: +49-251-70365-399
>>>> Email: vlad.cojocaru[at]mpi-muenster.mpg.de
>>>> http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
>>>>
>>>>
>>>>
>>>>
>>>> _______________________________________________
>>>> AMBER mailing list
>>>> AMBER.ambermd.org
>>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>>
>>> _______________________________________________
>>> AMBER mailing list
>>> AMBER.ambermd.org
>>> http://lists.ambermd.org/mailman/listinfo/amber
>>>
>>>
>> --
>> Dr. Vlad Cojocaru
>> Max Planck Institute for Molecular Biomedicine
>> Department of Cell and Developmental Biology
>> Röntgenstrasse 20, 48149 Münster, Germany
>> Tel: +49-251-70365-324; Fax: +49-251-70365-399
>> Email: vlad.cojocaru[at]mpi-muenster.mpg.de
>> http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
>>
>>
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>>
>
>

-- 
Dr. Vlad Cojocaru
Max Planck Institute for Molecular Biomedicine
Department of Cell and Developmental Biology
Röntgenstrasse 20, 48149 Münster, Germany
Tel: +49-251-70365-324; Fax: +49-251-70365-399
Email: vlad.cojocaru[at]mpi-muenster.mpg.de
http://www.mpi-muenster.mpg.de/research/teams/groups/rgcojocaru
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Sep 11 2012 - 07:30:11 PDT
Custom Search