Position restraints for Umbrella sampling simulation of membrane protein-ligand

Dear gromacs users,

I am planning to perform an Umbrella sampling simulation for a ligand translocating through a membrane protein. I am using pull-coordinate = direction and pull-vector = 0 0 1 to steer the ligand through the protein along the z-direction. In a gromacs tutorial by Dr. Justin Lemkul, I saw that they are applying position restraint on one of the pull-groups. Is it necessary to position restrain the membrane-protein that I am using for steering the ligand through the protein in addition to the pull-force? Any help is much appreciated thank you

Since your pull coordinate is a direction and you are pulling a ligand through the protein, you would like the protein to be restrained in the pulling dimension (at least the backbone of the protein). Otherwise, if it is “difficult” to pull the ligand through, the protein might follow the ligand.

Thank you for your reply. I have initially run a long steered MD simulation without position restraining the protein. The protein is like a beta-barrel structure embedded in a lipid bilayer. I am using the following pull code parameters for running a steered MD simulation first:

pull = yes
pull-ncoords = 1 ; only one reaction coordinate
pull-ngroups = 2 ; two groups defining one reaction coordinate
pull-group1_name = LIG
pull-group2_name = SOLU
pull-coord1_type = umbrella ; harmonic potential
pull-coord1_geometry = direction
pull-coord1_dim = Y Y Y ; as 0 0 1 already allows z-direction movement only
pull-coord1_groups = 1 2 ; groups 1 (Chain A) and 2 (Chain B) define the reaction coordinate
pull-coord1_start = yes ; define initial COM distance > 0
pull-coord1_rate = 0.0001 ; 0.0001 nm per ps = 0.1 nm per ns
pull-coord1_k = 100 ; kJ mol^-1 nm^-2
pull-nstxout = 500
pull-coord1-vec = 0 0 1

I checked and the ligand is translocating from one end to the other end. The protein is also rotating or tilting a little, but not moving away (most probably) as it is pretty difficult to observe. I have not yet used any position restraints on the protein. Do you think that using position restraints on protein backbone or C-alpha would be more appropriate in this case?

Is the order of the pull-groups actually matter anyway while we are pulling along the reaction coordinate as nothing related to that is mentioned in the manual?

In the current setting I could see the ligand moving along the z-direction. Earlier I have tried with pull-coord1-geometry = distance and I could not pull the ligand through the protein.
Thanks in advance

Do I understand correctly, SOLU is your protein?

Yes, the order of the pull group matters. In pull-coord1_groups the first entry is the reference group. So, I expect (but may be wrong) that you are actually pulling your protein (and presumable indirectly also the lipid bilayer) towards your ligand. I then assume that the reason why it looks like your are pulling the ligand through the protein is that you have center of mass motion removal, which means that the overall translational motion of the system will be removed, resulting in a net movement of your ligand in the opposite direction of your pull force.

Try without restraining the protein. It’s quite possible that it will work fine. Otherwise I’d recommend to just add enough restraints to keep it in place.

Thank you for your reply. The “SOLU” group refers to the Protein atoms. I was not aware of the importance of the order of the pull group as it is not explicitly mentioned in the gromacs manual. My aim is to run umbrella sampling simulations on configurations extracted from steered MD simulation. I have performed a set of umbrella sampling simulations by extracting configurations from the previously mentioned protocol (without using position restraints for the protein). I just checked the steered MD trajectory and it looks like the ligand is moving from one side to the other, but the lipid bilayer and the protein is slowly moving but not getting pulled towards the ligand, rather the ligand is getting pulled towards the protein.

So, if I select group1 = Protein and group2 = LIG, then ideally the ligand should be pulled with respect to the protein. Is that correct? And if I do not use a position restraint in this case, will that work correctly?

Also, if we generate the necessary configurations of the ligand sampling different locations in and around the protein core, does the order of the pull-groups matter during umbrella sampling simulations as we don’t pull the ligand during umbrella sampling anymore?

I should correct myself. The pull force is applied on both groups (in opposite directions). The first group is still the reference group.