Re: Accelerated Molecular Dynamics

From: Vijay Vammi (vsvammi_at_iastate.edu)
Date: Tue Oct 25 2011 - 23:45:32 CDT

Hi Yi,

Thinking more about it, increasing the omega dihedral spring constraint may
be the only way to handle this in NAMD. As the extra bonds feature does not
support sinusodial function, it would spike up if there was a change in the
sign of omega. (-180 and 180 are same).

But its not clear of the impact of spring constant on the sampling. Any
other way that could work?
Thanks
Santhosh

On Tue, Oct 25, 2011 at 9:04 PM, Vijay Vammi <vsvammi_at_iastate.edu> wrote:

> Hi Yi,
>
> Thanks for the reply. I like the idea of adding additional restraints to
> the omega dihedrals.
>
> Since I am interested in protecting all the omega diherdals, is it wise to
> increase the Kdihedral(may be double or triple the original value) in the
> parameter file for the atom types?
> I am right now doing a manual changie of the K for different combinations
> that I observe from the psf file for the omega angle.
>
> Is there a better way ? I did notice cispeptide plugin in VMD which gives
> a restraint file but I thought this might be a good solution.
>
> Thanks
> Santhosh
>
>
> On Tue, Oct 25, 2011 at 6:03 PM, Yi Wang <yiwang198_at_gmail.com> wrote:
>
>> Hi Santhosh,
>>
>> Currently the aMD implementation in NAMD can't let you selectively
>> apply the boost to certain dihedrals, and from the implementation side,
>> it's actually quite tricky how to do this effectively---we did give it some
>> thought but didn't proceed due to the substantial amount of time the coding
>> might require.
>>
>> An alternative at this point is perhaps to add some restraints to
>> protect those omega dihedrals being distorted by the boost?
>>
>> Best,
>> Yi
>>
>> On Oct 25, 2011, at 2:11 PM, Vijay Vammi wrote:
>>
>> > Hi,
>> >
>> > I have been using NAMD's implementation of Accelerated Molecular
>> dynamics for my research. Thanks for the implementation of the feature in
>> NAMD.
>> >
>> > In the original paper of AMD, there was no discrimination between the
>> different torsional angles (phi, psi, omega or chis). And the boost was
>> provided to all the dihedrals when required.
>> > While running AMD simulations I have observed that the Omega torsional
>> angle does get distorted (standard geometry being near 180) to values close
>> to 140 sometimes which is highly strained omega.
>> >
>> > Would there be any functional difference if dihedral potential
>> concerning Omega was not boosted or scaled to a value lower than
>> phi/psi/chi boost? Is it possible to implement such a feature into
>> NAMD(local for testing) without affecting the performance? Hope I was clear
>> in my question.
>> >
>> > Thanks
>> > Santhosh
>>
>>
>

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:20:56 CST