Re: queries about "Tclforces" utility in NAMD

From: harish vashisth (
Date: Sun May 27 2012 - 17:39:01 CDT

Dear Jerome,
 Thanks for your earlier suggestions. You were right; there is no issue so
far in using *colvars* and *tclforces* together. Also, I could define an
"orientation" colvar component to prevent my protein from rotating by
harmonic restraints (script below). I also wanted to stop translation of
the center-of-mass of my protein and activated "centerReference" keyword by
giving initial PDB file with B-columns of protein atoms tagged 1.0 (also
shown in colvar script below). But I still have a minor gripe. When i plot
the distance of my instantaneous COM of protein from COM at frame 0 of run,
i see a net drift in COM position which increases to 1.6 A and then
fluctuates back to lower values like 0.4 A, 0.5 A etc.. Can you possibly
spot any reason why this might be the case; also is there a better way
using colvar to stop the COM drift such as defining a dummy atom at initial
value of COM and tethering instantaneous COM of protein to dummy atom
position or something. Any suggestions would be helpful. Thanks a lot.

colvarsTrajFrequency 1000 # output values every 100 steps

colvar {
     name fix-rotation

     orientation {
           atoms {
                atomsFile ref_orient.pdb
                atomsCol B
                atomsColValue 1.0
                centerReference on
                rotateReference on
                refPositionsFile ref_orient.pdb
                refPositionsCol B
                refPositionsColValue 1.0
           refPositionsFile ref_orient.pdb
           refPositionsCol B
           refPositionsColValue 1.0


harmonic {
    name my_harm
    colvars fix-rotation
    centers (1.0, 0.0, 0.0, 0.0)
    forceConstant 500.0


On Fri, May 25, 2012 at 3:48 AM, Jérôme Hénin <>wrote:

> Hi Harish,
> On 25 May 2012 07:06, harish vashisth <> wrote:
> > Dear all,
> > I've been trying to use "tclforces" utility to add some external
> > forces to a protein. I have used tclforces successfully in the past, but
> > have following questions for what I am trying to carry out now:
> >
> > (a) Under tclforces, do we have access to transformation matrix routines
> > similar to vmd, which can be used on-the-fly to align particular atoms
> in an
> > NAMD run? Specifically, my protein diffuses during the simulation and the
> > way I am applying forces, either I have to remove net
> translation/rotation
> > of my protein in a running simulation or if I want to skip that, I need
> to
> > align a target structure on-the-fly to compute my force direction vector.
> These are not part of NAMD.
> > (b) related to (a)--is it possible to use "measure fit" vmd command
> under
> > tclforces. I guess not as vmd routines have no direct relation to NAMD's
> > source unless someone has implemented them under tcl-interpreter of NAMD
> > similar to commands like vecsub, vecadd etc.
> Same thing. Note that the colvars module does these things
> transparently for you, if there is any chance that your custom forces
> can be implemented there.
> > (c) it also appears that NAMD's tcl-interpreter does not understand tcl
> > commands: lassign, lvarpop...or am I missing something?..I have tried
> these
> > things on a serial binary of version 2.8/2.9..
> These commands are relatively recent additions to Tcl. If you manage
> to compile NAMD linked to a recent Tcl implementation (ie version
> 8.5), you'll have access to them.
> Best,
> Jerome

This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:34 CST