[gmx-users] COM pulling with gromacs 5.0.x

Romain Gautier IPMC gautier at ipmc.cnrs.fr
Tue Aug 22 09:45:04 CEST 2017

I have a problem to use the COM pulling with the Gromacs version 5.0.X 
or 5.1.x.
We are working on membrane simulations (using Coarse-grained Martini 
Force Field).
 From a big membrane system with POPC lipid, we selected a small patch 
of lipids in the center (POPX) and pulled this patch in z direction with 
a constant-force to create tubules.
I simulated using NPT with T-coupling and P-coupling (semiisotropic), to 
adapt the system to the pulling.
With Gromacs 4.5.X, it works fine (and we have already published some 
results with this method).
We used these COM pulling parameters:
; COM pulling
pull                     = constant_force
pull_geometry            = direction
pull_nstxout             = 10
pull_nstfout             = 10
pull_ngroups             = 1
pull_group1              = POPX
pull_group0              = W
pull_vec1                = 0 0 1
pull_k1                  = -300

I understand the protocol: number of group = 1 (only POPX will be pulled)

With Gromacs 5.0.X, I have problems with the new protocol because I want 
to pull only on one group (POPX, my lipid patch) in one direction (z 
axis) with constant force. We found after a lot of tries only these COM 
pulling parameters that work:
; COM pulling  gr 5.0.2
pull                = constant-force
pull_geometry        = direction
pull_dim            = N N Y
pull_nstxout        = 100
pull_nstfout        = 100
pull_ngroups        = 2
pull_ncoords        = 1
pull_group1_name      = POPC_W
pull_group2_name      = POPX
pull_coord1_groups    = 1 2
pull_coord1_vec       = 0 0 1
pull_coord1_k         = -300

I didn’t understand very well the new parameters because we have 2 
groups (pull_ngroups) not only one.
If I change the group 1 name with only W (waters molecules as with 
Gromacs 4.5.x), I have the same error message (during the tpr generation 
or during the first steps of simulation):

Fatal error:
Distance between pull groups 2 and 1 (50.142342 nm) is larger than 0.49 
times the box size (40.694592).
You might want to consider using "pull-geometry = direction-periodic" 

Why the distance is implicated in this case, I did not need distance 
parameters because I choice to pull in one direction using a constant 
force, not to pull using a distance between 2 groups?
If I use "pull-geometry = direction-periodic", the system is not in NPT 
condition and I want to use this condition to adapt the membrane to the 
pulling (x-y decrease with z increase).

I did not find clear explanations in the Gromacs manual nor in Gromacs 
tutorial (for example this tutorial: 
Any person has a solution and explanation?
Dr Romain Gautier

Cet email a fait l'objet d'une analyse antivirus par AVG.

More information about the gromacs.org_gmx-users mailing list