[gmx-users] pull code: distance between pull grp and ref grp is more than usual at the begining of simulation (at 1st time frame 1 ps)

Justin A. Lemkul jalemkul at vt.edu
Tue Sep 13 16:00:32 CEST 2011

Shilpi Chaurasia wrote:
> Dear Justin,
> Thanks a lot for your reply.
> Here I am providing the data and explaining the problem in detail:
> I hope that it will help you in understanding in a better way. 
> Considering the tetramer as 'abcd'.  'ab' as one dimer and 'cd' as 
> another, I want to separate 'ab' and 'cd'. Length of each dimer is 8 nm, 
> so the tetramer (initial configuration) is approximately 16 nm and I 
> have used a cubic box of 30 nm. Two groups (reference group and pull 
> group) are decided on the basis of residue numbers forming 'ab' as 
> reference grp and 'bc' as pull group by making an index file.

I will assume you mean CD here.  If your groups are AB and BC, then you'll 
certainly have problems.

> I have verified that the initial configuration does not split across 
> periodic boundaries, it is always inside the box throughout the 
> simulation. Moreover, the initial distance is not equal to box vector.
> By g_dist , I have tried to calculate the distance between the two 
> groups and it came out to be approx. 15 nm (Please have a look at the 
> dist.xvg file at the end). Do you think that in the first ps time frame, 
> should it get separated by 15 nm?? Please correct me if I have done 
> anything wrong.

If the dimer truly is together, then I would think this distance is too large 
unless (1) it actually is actually split across PBC, (2) your size calculations 
are wrong, or (3) your index groups are wrong.  In any case, I can see this 
being a problem.  A distance of 15 nm is 50% of the box size, and as soon as 
separation occurs, your pulling reference distance will be greater than 50% of 
the box vector, which causes a periodic distance to be calculated.  Either use 
the distance_periodic pulling geometry or use a larger box.

See below for a few thoughts on the dist.xvg data.

> By mistake, in my last mail I wrote that I need to pull along Y axis. In 
> actual, it is Z-axis and I have used pull_dim N N Y. So it is just the 
> mistake in writing the mail.
> Please let me know do you need any other data or parameters that I have 
> used during this simulation.

It would also be useful to know what grompp printed as the reference distance at 
time t=0.

Here's why I think periodicity is indeed an issue:

>    0.0000000   15.0342455   -0.7714367   -0.9718266   14.9829559
>    0.5000000   15.0345469   -0.7726345   -0.9766102  -14.9828863

Note that within 0.5 ps, the sign of the z-distance has changed but the 
magnitude is basically the same.  This indicates to me that your dimer is 
dancing back and forth across a periodic boundary.  The output of pullx.xvg 
would also be informative to confirm this phenomenon.



Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080


More information about the gromacs.org_gmx-users mailing list