Tutorial of sample umbrella

GROMACS version:2020.1
GROMACS modification: No
I was learning this tutorial:

Umbrella Sampling.

[quote=“Gray, post:1, topic:1494, full:true”]
GROMACS version:2020.1
GROMACS modification: No
I was learning this tutorial:

Umbrella Sampling.

When I did this step:

gmx grompp -f npt_umbrella.mdp -c conf6.gro -p topol.top -r conf6.gro -n index.ndx -o npt0.tpr

gmx grompp -f npt_umbrella.mdp -c conf449.gro -p topol.top -r conf449.gro -n index.ndx -o npt22.tpr

I got this error:

When the maximum distance from a pull group reference atom to other atoms
in the group is larger than 0.5 times half the box size a centrally
placed atom should be chosen as pbcatom. Pull group 1 is larger than that
and does not have a specific atom selected as reference atom.

Maybe I should expand the size of the cell, but why the tutorial doesn’t exist this problem?
Thanks for the advice of you.


Try to reproduce the tutorial using the GROMACS version for which it is designed, 2018. There may be differences in 2020 and I haven’t had time to look into what those might be.

Please don’t do this. The users of this forum give of their free time (what is that?) to help others. It may take a few hours or even a few days for someone to provide input.

Thank you for your advice. I will try it.

Sorry, my apologize.I won’t do this again.

Hello Gray,
I am new to the Gromacs and I am using Gromacs 2021.7. I am facing the same issue. How did you resolve it? With different systems, did the tutorial work?

I would like to add that trying to reproduce that tutorial with GROMACS 2024.1 also throws the same error originally reported by Gray in this thread.


If I remember correctly the tutorial defines the two pull-groups as the two chains of a protein.

pull_group1_name        = Chain_A
pull_group2_name        = Chain_B

You should be able to solve this error by specifying the pbc atom with the following keyword

pull-group1-pbcatom = ATOM_NUM

and by substituting ATOM_NUM with the number of the atom of the group Chain_A that is more or less at the centre of the group. You can look for such an atom by visualising your system (e.g. with VMD) and picking an atom that is ca. at the geometrical centre. On a side note, it is a bit weird that you are getting this error, I remember going through the tutorial and never encountering such a problem. I was likely using GROMACS2019.X, though.

Two things:

  1. A bit of a timeline here:

The tutorial was created using a version of GROMACS circa 4.0.2. (More than a decade ago.)

@obZehn seems to recall that version 2019.X completed without throwing that error.

@Gray reported in this thread that version 2020.1 throws the error.

@juhidutta10 reports to this thread that version 2021.7 throws the error.

I am reporting that version 2024.1 throws the error as well.

In tabular form:

version Event
4.0.2 (~2009) No event
2019.X No event
2020.1 Error
2021.7 Error
2024.1 Error

Looking at the table above, I see a dependency between version and errors. Don’t you?

  1. Also, version 2024.1 automatically picks a central atom to use as a reference for the centre of mass. In the case of the tutorial, chain A has 226 atoms and GROMACS automatically picked atom 113 when I explored it.

For that reason, when I tried pull-group1-pbcatom=113 in the .mdp, it made no difference and the error persisted.

What did help, a little bit, was to use pull-pbc-ref-prev-step-com=yes. That allowed the sampling to continue a few more configurations until the ligand transitioned from fold to a sort of lonely spaghetti and the error started to crash the simulation again.

At this point, I think that knowing from the start that a box much bigger than the one shown in the tutorial is required is the best solution to reduce time lost and frustration from the life of scientists. (I hope that explains why I invest time in providing these detailed responses.)


I think the central atom that is picked by GROMACS is just central in terms of number of atom, that is, if your group has 100 atoms then the number 50 is picked. This doesn’t ensure you that that is a reasonable geometrical centre of the group, which is why you should pick with that keyword an atom that represents reasonably the centre.

It seems to me that the problem is related to the chain extending and (the chain group) becoming larger than half the box size, which can generate problems in COM calculation with PBCs. I think you may want to both keep pull-pbc-ref-prev-step-com=yes and to pick pull-group1-pbcatom an atom that is around the centre of the group when it is extended. Take a look at one of the last frames and pick from there! Not sure it will work, but supposedly this should be well defined both in the starting compact configuration and in the more elongated ones. This also means that with a larger box you shouldn’t get this error, but you will need many more water molecules to solvate the system and, as such, you will most likely have quite an important performance drop.