Umbrella sampling for pulling one atom out of a big crystalline structure

GROMACS version: 2022.2
GROMACS modification: Yes/No
Here post your question

I am trying to pull one atom(Al20) out of a large crystalline structure(MOF_Al20, without atom Al20).
Step 1- pull the atom and get configurations
Pull code
pull = yes
pull_ncoords = 1
pull_ngroups = 2
pull_group1_name = Al20
pull_group2_name = MOF_Al20
pull_coord1_type = umbrella
pull_coord1_geometry = distance
pull_coord1_groups = 1 2
pull_coord1_rate = 0.05
pull_coord1_k = 1500
pull-group2-pbcatom = 6385
pull-pbc-ref-prev-step-com = yes

This step looks like no problem. Then I use the following parameters for nvt-equilibrium and umbrella sampling. However, the umbrella histograms after wham analysis are randomly distributed, not along the reaction coordinate. I do not know what is wrong since the configurations were checked visually with VMD and looked okay.

NVT and Umbrella sampling code:
Pull code
pull = yes
pull_ncoords = 1
pull_ngroups = 2
pull_group1_name = Al20
pull_group2_name = MOF_Al20
pull_coord1_type = umbrella
pull_coord1_geometry = distance
pull_coord1_groups = 1 2
pull_coord1_rate = 0.05
pull_coord1_k = 1500
pull-group2-pbcatom = 6385
pull-pbc-ref-prev-step-com = yes
pull-coord1-init = (value of the distance of the tested configuration and starting configuration).

Could someone help me check if my running parameters look okay or not?

By the way, can anyone give me some advice about using “pull-coord1-type = constraint” in my case? Maybe this one is more proper for my case?

Thanks a lot!

Best,
Xiaoli

For umbrella sampling you should set the rate to zero.

I don’t think you can use pull type constraint in your case because of technical limitations.

Thank you, Hess.
I have realized the mistake and changed the rate to zero.
May I ask why I can not the pull type constraint?
If I can define an atom as the center of mass of my big crystalline structure and then apply force to the pulling atom and the big structure, is it possible?

Thanks a lot for helping me.

Technically you can do that, but you get much more stable results when you pull against a large reference group. With a single atom as reference there is a risk for local distortions.