Date: Fri, 8 Nov 2002 12:50:42 -0800 (PST)
From: Thomas Cheatham
Subject: Re: residue cross-correlation


> I ask this question again because nobody helps me to answer
> it until now. Does anyone know how to analyse residue
> cross-correlation over a trajectory in Amber? I *really* need this
> analysis tool. If Amber can't do, please tell me how to do it using
> another package.

The likely reason that you haven't heard anything is that there is no
facility distributed in AMBER, currently, to my knowlege, that can do this
type of residue cross-correlation (sometimes called DCCM, dynamical
cross-correlation maps).

You might be able to get this out of CHARMM (by writing a complicated
script with the correl facility). It is also available in the MD
Toolchest developed at Wesleyan

http://ludwig.chem.wesleyan.edu/~wwang/mdtool

and is likely available in other codes. Look to see what methods/codes
various papers discussing this method used to calculate the residue
cross-correlation.

It would not be difficult to add this functionality to the ptraj module of
AMBER which is designed to be extensible and has a fairly detailed set of
comments in "actions.c" describing explicitly how to add new functionality
for processing trajectories.

Note that the analysis of the MD trajectory data (perhaps along with the
development of new force fields) is actually the most difficult and time
consuming part of MD simulation; actually setting up and running the
calculations is fairly straightforward. While we are striving to make
more and more analysis tools available, most of the tools you see have
evolved from a few people willing to share their code and put the tools in
a common environment. The tool development is application or research
driven, i.e. we develop the tools we need for the job at hand. In most
groups using various biomolecular simulation tools, most people tend to
develop their own tools for their own purposes. Many times, people
reinvent the wheel, but through this process they come to understand
exactly what analysis they are doing and further understand the nuances
and limitations in the code. So, if you want DCCM, I would suggest (1)
waiting to see from the e-mail reflector if someone out there has a DCCM
code they are will to share (you might also try the CCL list), (2) wait
for the AMBER developers to implement it (don't hold your breath), (3) or
sit down and write up the code yourself.

Good luck,