[gmx-users] free energy problem

[Berk Hess]

Hi,

You seem to be trying to decouple a 9 residue peptide from the rest of the system.
This "brute-force" approach will only work when you have an enormous amount
of computer power and you will have to tune all your parameters carefully.
You should consider if this approach is feasible.

You crashes are probably due to the fact that you turn off all non-bonded interactions
in the protein. That leads to a strange state which will be difficult to integrate.
If you really want to pursue this approach, you should not decouple the intra-molecular
interactions.
This is far simpler in 4.0 than what you have down now.
Simply use the original itp file and set couple-moltype in the mdp file.
You should also use an intermediate state with zero charges but full LJ interactions,
that will make the path much smoother.

Berk

From: ecino at uwo.ca
To: gmx-users at gromacs.org
Date: Thu, 17 Sep 2009 11:11:55 -0400
Subject: [gmx-users] free energy problem

Hello. I am trying to run free energy calculations on a protein - protein complex. I believe that I have made all of the necessary changes to my topology, and the nonbonded parameters are set to 0 for state B. I am using the GROMOS96 53a6 ff. My system runs fine with free energy turned off, but gives many lincs warnings when it is turned on, and lambda = 0 to 0.5. Ultimately, the run crashes. One thing that I can think of that may be a problem is that state B has a nonzero charge, while A is 0. I have attached my run input, and topology. Any help is very much appreciated. Thanks. 

 Elio Cino
 		 	   		  
_________________________________________________________________
See all the ways you can stay connected to friends and family
http://www.microsoft.com/windows/windowslive/default.aspx
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20090918/41c00c5c/attachment.html>