AMBER Archive (2008)

Subject: AMBER: energy problems in simulation with restraints

From: Daniel Cappel (daniel.cappel_at_uni-wuerzburg.de)
Date: Mon Mar 31 2008 - 04:39:13 CDT

Hello Amber-List!

I would like to simulate (Amber 9) a protein system in explicit solvent
with some of the protein atoms restraint to their initial positions.
During the first part of the simulation I want to gradually remove this
restraints. Since, I did not find a way to decrease the strength of the
restraining force automatically, I "by hand" performed different
simulation steps with different values for RESTRAINT_WT.

Due to this procedure two problems occurred:
If I start with a restraint weight of 1 kcal/mol and decrease it in
steps of 0.1, at the first change of the restraining force constant a
sudden change of the restraint energy from about 20 to 2000 and a sudden
conformational change comes along.
If I start with a higher value of 10 kcal/mol and decrease it in steps
of 1 kcal/mol the simulation crashes after the first change in the
RESTRAINT_WT with "vlimit exceeded". If I insert a short minimization
(only 50 steps) between two MD runs with different restraint weights,
EEL changes from about -100,000 to -1,000,000 during the simulation. On
the other hand the VDWAALS term increases. Then, this leads to a new
"vlimit exceeded" and "SANDER BOMB" in the following MD step.

Well, I should add, that a simulation of exactly the same protein system
without any restraining forces works fine.

What have I done wrong? Is the described procedure not OK? Why is the
electrostatic and the van der Waals energy changing so enormously during
a quite short minimization?

Any suggestions are welcome. Thanks in advance.
Regards
Daniel 

-- 

Dipl.-Chem. Daniel Cappel

| Institut fuer Pharmazie, Universitaet Wuerzburg |
| Am Hubland, D-97074 Wuerzburg, Germany |
| Tel. +49 (931) 888-5485 | Fax +49 (931) 888-5494 |
| http://www.pharmazie.uni-wuerzburg.de |

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu