[gmx-users] Assembling a good simulation starting point

Ran Friedman, Biochemisches Inst. r.friedman at bioc.uzh.ch
Fri Feb 19 10:21:45 CET 2010

Hello John,

How large was the force after EM? Large forces often results in systems that 
explode during the simulation. Also, did you minimise with or without 
solvent? You assessment seems correct - the initial structure wasn't 
minimised and the tools of the trade are trying different conformations, 
minimising in vacuo at first and using other modelling tools before Gromacs. 
Also, read a bit in the mailing list and search the literature for similar 

Good luck,

On Thu, 18 Feb 2010 19:08:11 -0800 (PST)
  John Ladasky <blind.watchmaker at yahoo.com> wrote:
> Hello everyone,
> I'm a fairly new GROMACS user.  I'm running GROMACS 4.0.5 on top of Ubuntu 
>Linux 9.10.  I am still learning a lot.
> I just tried to set up my first fairly complex simulation, and it failed.  
>I have a protein of interest, a beta barrel with a hydrophobic, 
>ligand-binding interior.  I am interested in making mutations to this 
>protein, with the goal of getting it to bind to a rather different ligand 
>than the one it normally binds.
> The way that I propose to go about studying this problem is to construct a 
>partially-unfolded version of the protein structure, add my ligand of 
>interest, and then run an energy minimization.
> My first naive attempt to construct the partially-unfolded protein was not 
>successful.  I knew that it might have problems, but I tried it anyway.  
>Using Biopython, I rotated the atomic coordinates so that the beta barrel 
>was parallel to an axis.  Then I simply pulled all of the atoms 3 Angstroms 
>away from the axis.  Finally, I inserted my ligand.  Visually, inspecting 
>the starting structure with PyMol, I didn't see anything egregious.  
>However, I could have some unwanted close contacts.
> I got a few "long bond" warnings from pdb2gmx, but I persisted.  I got 
>through genbox, editconf, and my first grompp sucessfully. But then when I 
>tried the first, position-restrained energy minimization, it aborted with 
>too many LINCS warnings.   I blew the system up.
> These LINCS warnings could come from close contacts, or from large forces 
>in over-stretched bonds which resulted from my crude approach to expanding 
>the protein structure.  Whatever the cause, I need a smarter way to start.  
>I am open to ANY suggestions!
> What I THINK I might want to do is to manipulate the starting structure in 
>a more natural way.  For example, selecting some peptide bonds in the beta 
>turns, and changing their angles.  A program which allows me to manipulate 
>structures, and not just simulate natural forces, is what I think I need.  
> Surely, people who have used GROMACS will have faced problems simliar to 
>mine.  Thanks for your advice!
> John Ladasky

Ran Friedman
Postdoctoral Fellow
Computational Structural Biology Group (A. Caflisch)
Institute of Biochemistry
University of Zurich
Winterthurerstrasse 190
CH-8057 Zurich, Switzerland
Tel. +41-44-6355559
Skype: ran.friedman

More information about the gromacs.org_gmx-users mailing list