[gmx-users] problem with pull code

[Justin Lemkul]

On 10/13/12 3:17 PM, Sanku M wrote:
> Hi all,
>    I am trying to do an umbrella sampling of a fairly extended state of the protein along end-to-end distance of it. I am using a rectangular box ( elongated along x ) of dimension 20 nm X 8nm X 8 nm. I am doing the simulation in NVT ensemble. gromacs 4.5.4 is being used.
>
> But, even after running for 1 ns, the simulation crashes saying that
>
> Program mdrun_5mpi, VERSION 4.5.4
> Source code file: pull.c, line: 329
>
> Fatal error:
> Distance of pull group 1 (2.940209 nm) is larger than 0.49 times the box size (9.000000)
>
>
> But if I see the trajectory, it hardly shows any issue with seeing the PBC image . I am not sure how to avoid this issue.

The pull code is very sensitive to these motions.  Mathematically, I don't 
understand why mdrun is complaining, but your setup is prone to failure.  An 
extended protein in an elongated box will surely interact with its periodic 
image at some point.

> Should I make one of the endgroup as freeze group so that it does not get tilted.

I don't think you'll be able to freeze or restrain enough of the protein to make 
a difference that does not significantly impact the dynamics of the system.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================