Pull code and pull-group1-pbcatom

GROMACS version: 2020.4

Hi all,

I’ve been pulling pre-stretched DNA along the Z-direction through a pore in a solid membrane + some solvent. My pull code has been:

pull = yes
pull-coord1-type = umbrella
pull_group1_name = DNA
pull_coord1_start = yes
pull_coord1_geometry = direction-periodic
pull_ngroups = 1
pull_ncoords = 1
pull_coord1_groups = 0 1
pull_coord1_origin = 0 0 0 ;default
pull_coord1_vec = 0 0 -1
pull_coord1_rate = 2.0e-5
pull_coord1_k = 5000
pull-nstxout = 5000
pull-nstfout = 5000

At some my system changed a bit and now it is much longer in the Z-direction, so grompp issued a bunch of setup errors and the following had to be added to the pull code:

pull-group1-pbcatom = 5419
pull-pbc-ref-prev-step-com = yes

I thought it worked, but it is actually clunky. The pulling rate seems to oscillate and sometimes there are distinct jerking motions of the DNA along Z. My assumption was that 5419 should be a central atom in the DNA molecule (say, a phosphorus), close to the membrane at the start of the simulation. Is this correct, or is that PBC atom supposed to be in the membrane? Alternatively, is there a way to set this up so that it works the way it did with the original pull code (above)?

Many thanks!


1 Like

Bump. Is there absolutely anyone who could help with pull code?

Did you try to use pull-coord1-geometry = cylinder?
See here for more details

Best regards

I’m not new to this, so, yes, I did read the documentation and this portion of it is rather poorly written. So far, constant-force seems to be the only way.

Just to avoid confusion, my suggestion was to use pull-coord1-geometry = cylinder, not to change pull-coord1-type. pull-coord1-type = constant-force should work with pull-coord1-geometry = cylinder.
in the mdp file you have uploaded pull-coord1-type = umbrella (not constant-force). I guess you have tried both.
Best regards

Yeah, I only changed umbrella to constant-force. For my system, cylinder seems questionable in the same way direction-periodic is, because the COM position of a periodic vertically positioned DNA actually does not do anything but oscillate.
I wish I could try a bunch of options, but the system is so heavy that I need days before finding out if things work. Constant force seems fine, except now one really needs to adjust it to get a reasonable pulling rate. Thanks for the suggestion anyway!

I have a question on the options added to pull the pre-stretched system along z-direction.
I wonder why the following options were needed to simulate the system?

pull-group1-pbcatom = 5419
pull-pbc-ref-prev-step-com = yes

I don’t understand how they work.
I know that the option pull-group1-pbcatom specifies a reference atom. But I am unsure what is a reference atom and for what the atom is used.


The reference is to some sort of a “central” portion of the system that will help the software treat PBC. Unfortunately, the documentation pertaining to the entire pull code section is pretty terrible.

maybe you can open an issue on GROMACS GitLab on how to improve the pull-code documentation.