From: Jérôme Hénin (jhenin_at_ifr88.cnrs-mrs.fr)
Date: Sun Jan 08 2012 - 17:18:28 CST
Just one word of caution: check carefully if the positional restraints
that you apply cause overall drift of the simulation box. The colvars
make it non-trivial to define variables that depend on absolute atomic
coordinates, but that is intentional. The rationale is simple: the
physics in most NAMD simulations is translationally invariant, and in
practice, translational drift may occur for various reasons
well-covered on this list in the past. Typically, the most relevant
coordinates are defined in relative terms, or in "internal
coordinates" considering the whole simulation box.
You case might be an exception, but I feel that that is worth
insisting on once in a while.
On 9 January 2012 00:03, Michael Zimmermann <michaelz_at_iastate.edu> wrote:
> Thank you for the reply Axel.
> You are correct in evaluating that ambiguity - I do generally mean
> positional restraints (tho there is an accompanying change in
> conformation). My missinterpretation of the description of the consref
> parameter is to blame for not catching that it could be used in this way (I
> read it as the initial position would be taken from the reference file, not
> the "reference position"). I just ran a test simulation and it looks like
> this type of constraint will work.
> Thank you for your time.
> On Sun, Jan 8, 2012 at 3:19 PM, Axel Kohlmeyer <akohlmey_at_gmail.com> wrote:
>> On Sun, Jan 8, 2012 at 4:04 PM, Michael Zimmermann <michaelz_at_iastate.edu>
>> > Dear NAMD community,
>> > I have been reading the 2.8 user's guide for SMD, TMD, harmonic
>> > restraints,
>> > rmsd colvars, etc. as well as the mailing list archive for the best
>> > solution
>> > to the general problem of "I would like pull a structure to a reference
>> > conformation" subject to the following:
>> > 1) only a subset of the atoms need to significantly move in space
>> > 2) they need to achieve a precise final conformation (so not SMD)
>> > 3) the atoms that don't need to be directed can't be fixed in space -
>> > ideally they would be simply unconstrained (so, not TMD)
>> > I say not TMD because if I (1) only tag some atoms as TMD atoms, they
>> > will
>> > be aligned before the forces are calculated (thus no "pulling" through
>> > space
>> > will occur), or (2) tag all (or a sufficiently large subset) atoms as
>> > TMD,
>> > but now there are constraints added to the atoms that I would like to be
>> > unconstrained. The rmsd colvar routine also aligns prior to force
>> > calculation.
>> please clarify one detail. i don't quite understand
>> why it matters so much to you to not align the reference
>> to the current state, if you care about conformation.
>> conformation in my understanding are the relative
>> coordinates of a group of atoms. if you turn off this
>> alignment, you have what i would call "position restraints"
>> and thus your goal would be to direct atoms to specific
>> positions, and not only to a particular conformation.
>> so if i am right, then you should just do the opposite
>> of what people do to carefully equilibrate a system,
>> i.e. define position restrains and gradually increase
>> (not decrease) the force constant.
>> if you don't need to reach specific coordinates in space,
>> then you should explain where the difference lies.
>> > The most convenient solution would be a parameter that allows me to turn
>> > the
>> > alignment off in TMD or rmsd colvars. I have been unable to find such an
>> > option.
>> > To some extent, TMDDiffRMSD could accomplish this, but I don't want the
>> > structure to back-track to the first conformation.
>> > I am presently using a custom tclforces script to add forces
>> > specifically
>> > where I want them, but this code is difficult and time consuming (for
>> > me) to
>> > maintain. If there is a way to turn off the alignment in TMD (that is,
>> > prevent the TMD atoms from being spatially superimposed prior to force
>> > calculation), would you be so kind as to help me with the syntax? I have
>> > been unable to find something like it in the archive or User's Guide.
>> > Thank you,
>> > Michael T. Zimmermann
>> Dr. Axel Kohlmeyer
>> akohlmey_at_gmail.com http://goo.gl/1wk0
>> College of Science and Technology
>> Temple University, Philadelphia PA, USA.
> Michael T. Zimmermann
> Ph.D. in Bioinformatics and Computational Biology
> Department of Biochemistry, Biophysics and Molecular Biology
> Iowa State University
This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:06 CST