Date: Fri, 17 Mar 2000 08:21:15 -0800
From: David Case
Subject: Re: estimate entropy of dna using nmode

On Thu, Mar 16, 2000, Frank Yan wrote:

> we're using nmode to estimate the entropy of a dna sequence with a
> carcinogen binded. we took out the structure from dynamics trajectory, then
> do minimization exclude water and ions but using a 4r dd to mimic solvent.
> we found that after minimization using a drms of 0.0001 the structure is
> extremely distorted to a kind of denatured state...

You are seeing a big problem (perhaps an Achilles heel?) of the mm_pbsa
approach: there are no really good ways to estimate entropy changes for
many interesting types of transformations. This is still a very active
area of research.

If you want to use normal modes, you have to go to a very small gradient
in order to get useful results. Otherwise, you will have "imaginary"
frequencies whose contributions to fluctuations is not well-defined.

...good luck....dac