Re: FEP: FATAL ERROR: Periodic cell has become too small for original patch grid!

From: List (zheng042003_at_yahoo.com)
Date: Thu Mar 13 2008 - 16:16:04 CDT

Thank Jeff very much for the helpful comments.

1) Indeed the partially-solvated system shrinked by a few angstroms in some dimensions after FULL minimization. In the FEP simulation, since only the site of interest plus surrounding residues are allowed free to move with the remaining residues/atoms fixed, probably the SAME condition should be applied to the minimization prior to FEP: the peripheral atoms/residues should be fixed during the minimization. I'll try so to see if it solves the problem.

2) Actually I divided the full FEP into two jobs, one with lambda 0->0.5, and the other with lambda 0.5->1.0. Both FEP simulations start with the same minimized structure with the following settings:
 
        timestep 0.5
        fepEquilSteps 40000
        set nSteps 200000

 Interestingly, the second FEP is running fine (100,000 steps so far), while the first one crashed after running for 26,400 steps.

Best regards,

LIST

"Jeffrey J. Potoff" <jpotoff_at_chem1.eng.wayne.edu> wrote: What is happening is that your system volume has changed significantly
during the course of the simulation. Look for the following problems:

1. Instability due to overlap of atoms, or other highly unfavorable
interactions. This will generate an enormous force; system blows apart.

2. System isn't equilibrated. Initial density is too high or too low.
Either way, if the simulation has to go a long way to get to the "correct"
density, you will get this error. As long as the system isn't flying
apart, you can restart from the final configuration and eventually will
get something stable where NAMD doesn't complain about the patch grid.

Regards,
  Jeff

On Wed, 12 Mar 2008, List wrote:

> Dear NAMD experts,
>
> I have some questions about FEP (Free Energy Perturbation) and MCTI (Thermodynamic Integration). The configuration file I used is attached. Here are my questions:
>
> 1) I'm running FEP simulation in NPT ensemble. My system is a protein with a small water box enclosing the site of interest, and I'm running non-PBC FEP with atoms far away from that site fixed. According to some early posts in NAMD mailing list, I have the following settings:
> wrapAll off
> wrapNearest off
> PME no
> (All cellBasisVector1, ..., celOrigin lines have been removed)
> margin 2.5
> useFlexibleCell no
>
> But the FEP simulation CRASHED with the following error:
> **********************
> FATAL ERROR: Periodic cell has become too small for original patch grid!
> Possible solutions are to restart from a recent checkpoint,
> increase margin, or disable useFlexibleCell for liquid simulation.
> ************************
> I even tried to increase margin to a much bigger number but it didn't help. I'm not using PME/PBC, so I don't understand why it complains about the periodic cell. Could someone help?
>
> 2) I also tried NVE ensemble, but I got the message saying that FEP only works for NPT. Is this right?
>
> 3) In the FEP, I want to do thermodynamic integration, so I included the MCTI block in the configuration file AFTER the FEP specification. I wonder if this would work.
>
> Any help will be greatly appreciated.
>
> LIST
>
>
>
>
>
>
> ---------------------------------
> Be a better friend, newshound, and know-it-all with Yahoo! Mobile. Try it now.

======================================================================
Jeffrey J. Potoff jpotoff_at_chem1.eng.wayne.edu
Associate Professor Wayne State University
Department of Chemical Engineering and Materials Science
5050 Anthony Wayne Dr Phone:(313)577-9357
Detroit, MI 48202 Fax: (313)578-5815
http://potoff1.eng.wayne.edu
======================================================================

       
---------------------------------
Never miss a thing. Make Yahoo your homepage.

This archive was generated by hypermail 2.1.6 : Wed Feb 29 2012 - 15:47:38 CST