Re: use of colvars to restrain the absolute positions of all atoms in a large system

From: Jérôme Hénin (jerome.henin_at_ibpc.fr)
Date: Wed Jul 10 2013 - 05:07:16 CDT

Hi Abir,

You may obtain the same effect as positional restraints within the colvars by restraining a RMSD variable, for which you disable the least-squares fit process (thus calculating the raw, unfitted RMSD instead of the minimal RMSD). This, however, will incur a performance penalty compared with NAMD's built-in positional restraints ("harmonic constraints", which badly need to be renamed).

Conversely, there are ways you can calculate the restraint free energy, even for NAMD's built-in restraints, through post-processing of the trajectory: it is very easy to calculate the restraint forces on each atom, and then you have a choice of several estimators for the free energy.

Finally, given what you are trying to accomplish, I am not convinced that restraining all atoms in the system is a good idea.

Cheers,
Jerome

----- Original Message -----
> Hi Giacomo and Aron,
>
> Thanks for the suggestions.
>
> I am using the alchemical simulations implemented in NAMD to
> calculate the free energy difference for a particular
> transformation. However, due to hysteresis related issues, I need to
> perform this transformation in steps. One of the steps is to
> calculate the loss of configurational entropy associated with the
> process of putting these restraints on all the atoms. The next step
> would be to calculate the free energy change for a process within
> this restrained system.
>
> Hence, the reason for using the colvars (and not constraints) is
> because we would like to obtain the free energy of the process as we
> slowly remove the restraints.
>
> We will still be doing dynamics because we are restraining the atoms,
> and not constraining ( or freezing) them.
>
> In other words, the goal is to restrict the ribozyme to only local
> motions, and find out the cost related to that in terms of free
> energy.
>
> I hope I am not too confusing.
>
> Thanks,
>
> Abir
>
>
>
> On 07/09/2013 05:16 PM, Giacomo Fiorin wrote:
>
>
> Hello Abir, you can use the colvars module to restrain the center of
> mass of your protein to stay near the dummy, so that the protein
> doesn't drift while its internal dynamics is unaffected. This may be
> helpful for you to visualize the trajectory, but the FEP calculation
> doesn't require it.
>
>
> If you want to restrain each atom individually, that requires
> defining a variable for each atom, which in the end defeats the
> purpose of defining "collective" variables?
>
> The most efficient way to apply restraints atom-by-atom is still the
> "constraints" keyword: what is the reason why you can't use it?
>
>
> Giacomo
>
>
>
> On Tue, Jul 9, 2013 at 6:01 PM, Abir Ganguly <
> 7.someone.iitd_at_gmail.com > wrote:
>
>
> Hi NAMD users,
>
> I have a very specific question regarding the use of the colvars
> module that I am stating below.
>
> I am trying to use the colvars module to restrain the absolute
> positions of all the atoms in my system, which is a ribozyme. I am
> running some alchemical FEP simulations on the system and I need to
> fix(restrain) the positions of all these atoms during the
> simulations.
>
> I can not use the constraints option in NAMD because of reasons that
> I can state later, if required.
>
> I realize that I can define a dummy atom, and can use the 'distance'
> option in colvar to harmonically restrain the position of all atoms
> to that dummy atom. I was curious what would be a clean and smart
> way to go about it, since I have so many atoms. It will be really
> helpful if there are some templates with these sort of colvars, so
> that I can get some directions as to how to build my input file.
>
> Thanks,
>
> Abir
>
>
>
>

This archive was generated by hypermail 2.1.6 : Tue Dec 31 2013 - 23:23:27 CST