Date: Thu, 25 Apr 2002 13:30:24 -0500
From: David Pearlman
Subject: Re: PMF calculations

You have encountered a limitation of the implemention
of NCORC ("PMF bond contribution"). This arises from the
fact that to calculate the PMF bond contribution for a bond
whose length changes with lambda, we need to determine the
atoms that move at each end with a change in the bond length.
It is not possible to determine these atoms in the case of a bond
that is within a closed ring, e.g in the case of guanine to adenine.

You have three choices in this case: You can ignore these contributions
(NCORC=0, INTPRT=5), in which case your calculated energy will not
be formally correct, although it may be qualitatively indicative.

Or you can set up your system to perform the free energy change
using "dual topologies". This is a pain in the neck to set up
with Amber, but can be done and will avoid the problem. See Pearlman,
JPC 98, 1487 (1994) for more details on the issue of "dual topologies".

A third possibility is to determine the "PMF bond contributions" using what
is termed the "constraint force" method, by setting ITIMTH = 1. The constraint
force method can be used for bonds within a closed ring that are changing
with lambda. HOWEVER, if you use the constraint force method you will
also need to run a separate simulation to moment of inertia correction, which
is not performed by Gibbs (that is, you'll need to write the code to do
this calculation). For more on the constraint force method, see
Straatsma, Chem Phys Lett, 196 p.297 (1992) and for a comparison with the
potential force method used by default in Amber, see Pearlman, JCP 98 p8946 (1993).

dap

Sophia Kondratova wrote:
>
> Hello,
>
> I have a question concerning "PMF bond contribution method", which is used to
> calculate the intra-perturbed group interaction contributions to the free
> energy while doing FEP calculations (NCORC and INTPRT flag).
>
> What other parameters besides NCORC=1 and INTPRT=1,2,3,4 or 5 have to changed
> in order to calculate these intra-pert group contributions? I am trying to do
> an FEP of guanine to adenine and when I set NCORC=1, I get values of ~600
> kcal/mol, which is completely wrong (I do get the right answer with NCORC=0
> and INTPRT=3)
>
> Any help would be appreciated
>
> Sophia Kondratova
> Chemistry Department
> University of New Brunswick
> Canada