Re: rigid body simulation with NAMD

From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Fri Nov 29 2019 - 05:19:45 CST

Hi Jovan,

1) NAMD doesn't have the kind of constraint algorithms that would make this
cluster literally rigid. (I'm still waiting for the "constraints" option
to be renamed "positionRestraints", I might just publish a NAMD fork that
only does that as a form of protest.*)
2) you can achieve a physically correct result (reasonable rigidity, as you
said) using harmonic restraints, and then extrabonds is probably the most
convenient way.
3) then all you need is a trade-off between rigidity and stability of the
simulation. If you go much higher than the usual force constants for bonded
terms in the force fields, then you'll need to reduce your integration
timestep accordingly. So you'll need to ask yourself how much fluctuation
you can tolerate. My guess is, typical bonded force constants will give
fluctuations that are physically acceptable, given the level of
approximation of the force field overall.

Cheers,
Jerome

* Too bad that this would break everyone's workflow, including CHARMM-GUI.
Sometimes you have to do what you've gotta do.

On Fri, 29 Nov 2019 at 09:50, Jovan Dragelj <jovandragelj_at_gmail.com> wrote:

> Thanks for your answer.
> QM/MM is one way to describe this system considering the lack of
> parameters but the goal here is to run MD simulations of over 500ns
> followed by GaMD.
>
> My question was more in the direction of which of those methods I
> presented can be used to achieve reasonable rigidity of one part of the
> protein.
>
> On Thu, Nov 28, 2019 at 10:54 PM Victor Kwan <vkwan8_at_uwo.ca> wrote:
>
>> Have you considered QM/MM?
>>
>> Don't think NAMD have support for rigid bodies.
>>
>>
>>
>>
>>
>> On Thu, Nov 28, 2019 at 4:17 PM Jovan Dragelj <jovandragelj_at_gmail.com>
>> wrote:
>>
>>> Hi,
>>>
>>> I have problems finding a proper way to simulate rigid body with NAMD.
>>> The system is an enzyme with a metal cluster for which we do not have FF
>>> parameters and want it treated as a rigid body. Protein needs to be able to
>>> rotate fully, therefore *fixing *atoms won't help.
>>> Unfortunately, I could not make any simulation stable (let's assume
>>> other simulation conditions are fine, since we using relatively standard
>>> protocols).
>>>
>>> 1. I tried using *freeEnergy *feature with urestraint, by defining ref
>>> values from the crystal structure and high(er) constants, bond 500, angles
>>> 100, barrier for dihedrals 1.0
>>> 2. If using *extraBonds *feature, how is this approach any different
>>> than number 1.? I tried this with same values as in 1.
>>> 3. I have not tested this but, can this be achieved by using simply high
>>> *constraints* on atoms? Note: We do not have any FF parameters apart
>>> from partial charges.
>>>
>>> Details: CHARMM36 FF, NAMD 2.13
>>>
>>> Any help would be highly appreciated.
>>> Thanks,
>>> Jovan
>>>
>>
>
> --
> Dr. Jovan Dragelj
>
> AG Mroginski
> Biomolecular Modeling
> Institute of Chemistry
> Technical University Berlin
> Strasse 17. Juni 135
> 10623 Berlin
> e-mail: jovan.dragelj_at_tu-berlin.de <jovan.dragelj_at_campus.tu-berlin.de>
> Mobile: +49-176-27091982
> Tel(office): +49-30-314-26516
>

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2019 - 23:21:02 CST