Re: AMBER: TI - frcmod file

From: David A. Case <>
Date: Wed, 13 Jul 2005 14:10:56 -0700

On Wed, Jul 13, 2005, Ilyas Yildirim wrote:
> I am a little bit confused on one thing. Maybe someone can help me with
> this. Now, let's say we load the parameter files from parm99.dat first,
> and then the frcmod.03 file. First, the torsional parameters for CT-CT-C
> -N will be
> CT-CT-C -N 1 0.100 0.0 -4. phi,psi,parm94
> CT-CT-C -N 1 0.07 0.0 2. phi,psi,parm94
> Then, after the second parameter load (frcmod.03), what will be the
> torsional parameters of CT-CT-C -N? I thought that the last column
> (above, it is -4 and 2) is defining if there are multiple torsional
> parameters for a particular torsion angle. So, if we wanted to change
> them, we should have used the following format: (at least that's what I
> have understood)
> CT-CT-C -N 1 XXXXX XXX -4.
> But in frcmod.03 file, the torsional parameters for CT-CT-C -N are defined
> as following:
> CT-CT-C -N 1 0.0560 0.00 -3.
> CT-CT-C -N 1 0.0657 180.00 -2.
> CT-CT-C -N 1 0.7784 180.00 1.
> I am thinking that these extra three torsional parameters for CT-CT-C -N
> are going to be added on the CT-CT-C -N parameter list (because their PN
> number are different; -4,2,-3,-2,1). If I am wrong, please correct me.

Note that only the absolute value of PN matters: 2 is the same as -2.

> If it overwrites the parameter list, how does it do it? Just deletes the
> whole CT-CT-C -N torsional parameter list previously defined, and then adds
> the new parameter list?

When a frcmod file encounters a set of torsions (like CT-CT-C -N) it indeed
deletes the whole original set (which may have several Fourier components,
like 2 and 4 in this exmaple), then adds the new terms (which may also have
several components, 1,2 and 3 in this example). So you end up with just the
three components in the frcmod file, and none of the terms that were in the
original (parm99.dat) file.

In my view, to be fully safe for future code revisions, people who are making
frcmod files should explicitly override all existing Fourier components. In
this example, that would mean adding a line like this to frcmod.ff03:

CT-CT-C -N 1 0.0000 0.0 -4.

This would make it uambiguous that the four-fold term in ff03 was not really
there (==same as being there with a force constant of zero.) This would allow
simpler algorithms to be used, would guard against bugs (such as in the
distinct, but similar problem of overriding wild-card torsions with specific
torsions) and would reduce confusion among humans who look at these files.


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Wed Jul 13 2005 - 22:53:00 PDT
Custom Search