amber-developers: Junmei weighs in on torsions

From: David A. Case <>
Date: Wed, 29 Jun 2005 17:07:30 -0700

OK: Junmei has sent me the following mail:

> For those torsional parameters, I intended to use just one-fold term. As
> I remembered, the use of one and third folds made the fitting worse. As
> I understand, in Leap, one needs to put specific torsional terms behind
> the more general ones so that the specific ones will replace the generic
> ones when the parameter files are read in. I got the same results when I
> used the parm99.dat and the frcmods files.

So, my understanding was wrong, (although I still sort of agree with Tom's
comments on halogens.) This means that the "rule" is that a specific
term will replace a wild-card three-fold term. This is "good" in the
that it is what LEaP does under most circumstances, so we do not need to
about undo-ing (or redo-ing) a big body of work.

I've attached an updated "parm99.dat.new2", which puts in explicitly a
zero-force-constant three-fold term along with the one-fold term, to
(I hope!) the "copy" problem from getting confused. We should test this
see what it produces under various circumstances.

I will comment on the improper ordering examples Tom brought up in a
subsequent e-mail, once I can grok what PLEP is really doing. Also, there
is some other weird behavior that Tom pointed out that still needs
investigating, but I am hopeful that the attached file will get us closer.


Received on Wed Apr 05 2006 - 23:49:55 PDT
Custom Search