AMBER Archive (2008)

Subject: Re: AMBER: bond breaking in proteins

From: Gustavo Seabra (
Date: Wed Feb 27 2008 - 10:09:58 CST

Hi Eddie,

Looking at your input, I believe your problem is likely to come from
the choice of the QM method. Looking at your PDB file, I noticed that
the phosphate is attached to the nitrogen just by a Hydrogen bond, but
you chose to use MNDO (qmtheory=3) to treat your QM region, a method
that is unable to correctly describe hydrogen bonds. See:

      1. Burstein and Isaev, Theor Chim Acta, 64, 397-401 (1984)

Notice that the choice of the QM method to use is delicate. You need
to know your system, and you need to search the literature for the
most appropriate QM method. In your case, you have the PO3 connected
to the N by a H-bond, so you are trying to describe the energy of
breaking this H-bond. This is a classical challenge for semi-empirical
QM, and MNDO just can't do it unless it is corrected for that, which
the version in Amber isn't. One starting point would be this article:

      2. T. Clark, J. Mol. Struct (THEOCHEM), 530, 1-10 (2000)

Notice that AM1 also had problems with H-bonds, and that PM3 was
parameterized (among other things) to improve on this.

So, my final advice to you is that you first look at the literature to
see what has been done before, what semiempirical method works for
your system or similar things. Also, you can try small calculations
with different QM methods, to see what happens in each case, and pick
the method that best describes your problem.

The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to