AMBER Archive (2003)

Subject: Analysis of close waters

From: Markus Drumm (mdrumm_at_biomedicale.univ-paris5.fr)
Date: Wed Feb 26 2003 - 06:23:25 CST

I am trying to do an analysis of the hydration of a specific site of my
  Protein-DNA complex. I've got a 10ns dynamics run in explicit solvent
and what I want is the complete story of the water molecules at this
specific site. I've found several waters "coming and going" until
approx. 5.0 ns. I used the procedure described as an H-Bond example in
the carnal section of the manual of Amber6. I.e. in a first run, I ask
carnal for a distribution of the water molecules around my specific
site. So, i get a file with the closest distance for each water molecule
during the whole simulated time. I use awk to select those waters which
are very close and do a second carnal run with these selected waters to
get a distance vs time plot for each molecule.
As a whole, the procedure seems to work, but I get lots of false
positive results. i.e. In the first carnal.out (the distribution) a
water molecule is said to be at closest say 2.6 A away from my site. I
do the second run, plot the distance vs time, and it is nowwhere near my
site at any time, i.e 100 or more A away for the whole time. I
crosschecked several times my molecule numbers and there is no doubt.
Is this a known bug of the carnal DIST option? Has anybody carried out a
similar analysis on a comparable system (116 residues, 9900 waters)?
Which procedure do you use? for disk space reasons I cannot unzip my
whole trajectory with waters and look at it with vmd. But if the problem
persists, it's probably what I'll do 50ps per 50ps.
Thanks in advance for any comment,

Markus. 

-- 
********************************************************************************
Dr. Markus Drumm
Postdoctoral Research Fellow
Equipe de Chimie Bioinorganique et de Modelisation Moleculaire
CNRS UMR 8601
45, Rue des Saints-Peres
75270 Paris Cedex 06
France
********************************************************************************