[gmx-users] Re: Umbrella Sampling Pull code Problem

Sheeba sheeba.jem at gmail.com
Fri Aug 31 17:56:04 CEST 2012

As Bogdan suggested this could be an issue with the sign of the distance.
When you run grompp you get 

>Pull group  natoms  pbc atom  distance at start     reference at t=0
>       0       994       497 
>       1         2     46347   1.224                 1.950      

which suggests your starting structure is at 1.22 nm but it could be +/-
1.224 nm you might want to manually check the distance for your starting
structure. Gromacs restraints does not distinguish the sign of the distance
so when u started running the simulation, since you have a huge force
constant of 10000 kJ/mol, the pulling group is pulled to the window distance
of 1.95 except that it could be -1.95 instead of +1.95 resulting in a COM
distance around 2.7 nm. 

I had a similar problem when I tried pulling a peptide into the bilayer with
the reference group being the center of mass of the bilayer. For the window
at 0.5 nm, my starting structure had the pull group around -0.5 nm which I
did not notice was off by 1 nm since I had used g_dist to check the
distances. However the simulations ran fine with the pullx file listing
distances around -0.5 nm. This should not ideally happen if the restraining
section of gromacs distinguishes postive and negative distances because if
it did, then the peptide would have been pulled to the window distance of
0.5 nm within the first few ps of the simulation. I used 'pull_geometry =
distance' and I think gromacs restrains the pull_group at the magnitude of
the distance although the pullx file lists negative distances. My
understanding is that it will make no difference whether you give +1.95 or
-1.95 for the pull_init1 unless your starting conformations have the
distances at +1.95 and -1.95 respectively.  

Also another problem I had with negative distances is that g_wham does not
distinguish the sign of the distances so even if the pullx file has negative
values the pmf will be given only for the magnitude of the distances which
is not what you want. 

So it will be best if you can choose a reference group such that the
distance between your pull group and the reference is always positive. For
my case I used the center of mass of the phosphate atoms in the lower
leaflet of the bilayer as the reference group to pull the peptide into the
bilayer instead of using the center of mass of the bilayer. 

The important thing is also to manually check the distance between ur pull
and ref groups for the starting structures and start from a structure very
close to the window position especially with large force constants so you
can avoid sudden pulling for the pull groups.

Please note, I used Gromacs 4.0.5 for my umbrella sampling simulations and I
am not sure if what I observed is applicable to version 4.5.3.


Sheeba J. Irudayam
Postdoctoral Researcher
Department of Chemistry
University of North Carolina at Chapel Hill
View this message in context: http://gromacs.5086.n6.nabble.com/Umbrella-Sampling-Pull-code-Problem-tp5000715p5000725.html
Sent from the GROMACS Users Forum mailing list archive at Nabble.com.

More information about the gromacs.org_gmx-users mailing list