Subject: RE: AMBER: atom type CM
From: Junmei Wang
Date: Tue, 20 Jan 2004 10:43:23 -0600






Hi, Kristina,
For the torsional angles, the later appeared parameters override the
previous ones and this is reason why the generic torsional angle parameters
are placed before the specific ones in the force field files. The basic
philosophy in our force field development is to apply generic torsional
angle parameters and the specific ones are introduced only if they can
significantly improve the performance. Take -CM-CT- as an example,
according to the bond-antibond interaction theory, X-CM-CT-X is V2 with a
force constant of 0.0. However, a specific torsional angle HC-CT-CM-CM
makes the force field reproduce the experimental energies much better. You
may also try to apply CT-CT-CM-CM as well. However, we thought HC-CT-CM-CM
is more "general" than CT-CT-CM-CM. In the general amber force field
(gaff), we considered much more training set molecules and applied very
limited number of specific torsional angle parameters. Therefore, gaff
should have a much better parameter transferability.

Best

Junmei

Junmei-
Unfortunately, I had already generated parameters of my own for
non-conjugated
sp2 carbon before the CM epiphany... Now I want to make sure mine are
compatible with the existing force field, as well as being as transferrable
as
possible by making any reasonable adjustments based on the CM template.

Please forgive the detail--I've been thinking about parameters way too much
for way too long, as I'm sure you'll understand. I'm a little confused by
the
torsions involving CM. For torsions centered on -CM-CT-, there is a generic
X-CM-CT-X that is zeroed out, then a specific torsion for HC-CT-CM-CM with
two
terms, but nothing for other -CM-CT- torsions, CT-CT-CM-CM for instance.
Were
such torsions not present in the alkenes you were interested in or was
there a
reason to zero something like the CT-CT-CM-CM torsion and have HC-CT-CM-CM
carry the load?

X -CM-CT-X 6 0.00 0.0 3. JCC,7,(1986),230
HC-CT-CM-CM 1 0.38 180.0 -3. Junmei et al,
1999
HC-CT-CM-CM 1 1.15 0.0 1. Junmei et al,
1999

X -CM-CM-X 4 26.60 180.0 2. intrpol.bsd.on
C6H6
CT-CM-CM-CT 1 6.65 180.0 -2. Junmei et al,
1999
CT-CM-CM-CT 1 1.90 180.0 1. Junmei et al,
1999

At the same time there is a non-zero generic term for X-CM-CM-X, and
non-zero
specific terms as well for CT-CM-CM-CT. I must admit I'm not familiar with
how
AMBER combines generic and specific torsions, but I've generally been told
not
to mix the two types of parameters for the same central atoms. In this
case,
would the specific torsion override the generic for the torsion mentioned,
then the generic would be used for any other torsions involving the same
central atoms?

Thanks,
Kristina

>===== Original Message From Junmei Wang =====
>When we developed parm99 modeling some small molecules, Prof. Kollman
>suggested us to apply 'CM' for non-conjugated sp2 carbon in alkenes. This
>is the reason that there are some parameters with "Junmei et al, 1999" in
>the comments. Therefore, you may assign the non-conjugated sp2 carbon in
>alkenes to "CM" if you use parm99.
>
>Best
>
>Junmei
>