AWH accelerated sampling

GROMACS version:2020
GROMACS modification: Yes/No
Here post your question

Dear GROMACS community,

The complex I am trying to simulate has protein ( 10 chains), DNA, and RNA
I ran the simulation for 2 us so far but the biological event I want to observe requires a very long MD simulation. I decided to accelerate the sampling using the AWH method. I am not familiar enough with the AWH method or other methods such as umbrella sampling. I have watched the two webinars on AWH but I am still not sure about the mdp options.
Attached is the mdp I used and the error I got!
I believed that I need to choose a reference atom for pulling, But on what basis?

pull = yes ; The reaction coordinate (RC) is defined using pull coordinates.
pull-ngroups = 12 ; The number of atom groups needed to define the pull coordinate.
pull-ncoords = 7 ; Number of pull coordinates.
pull-nstxout = 1000 ; Step interval to output the coordinate values to the pullx.xvg.
pull-nstfout = 0 ; Step interval to output the applied force (skip here).

pull-group1-name = Protein_chain_A ; Name of pull group 1 corresponding to an entry in an index file.
pull-group2-name = Protein_chain_B ; Same, but for group 2.
pull-group3-name = Protein_chain_C
pull-group4-name = Protein_chain_D
pull-group5-name = Protein_chain_E
pull-group6-name = Protein_chain_F
pull-group7-name = Protein_chain_G
pull-group8-name = Protein_chain_H
pull-group9-name = Protein_chain_I
pull-group10-name = Protein_chain_J
pull-group11-name = RNA
pull-group12-name = DNA

pull-group1-pbcatom = 0
pull-group2-pbcatom = 0
pull-group3-pbcatom = 0
pull-group4-pbcatom = 0
pull-group5-pbcatom = 0
pull-group6-pbcatom = 0
pull-group7-pbcatom = 0
pull-group8-pbcatom = 0
pull-group9-pbcatom = 0
pull-group10-pbcatom = 0
pull-group11-pbcatom = 0
pull-group12-pbcatom = 0

pull-coord1-groups = 1 2 ; Which groups define coordinate 1? Here, groups 1 and 2.
pull-coord2-groups = 3 4
pull-coord3-groups = 5 6
pull-coord4-groups = 7 8
pull-coord5-groups = 9 10
pull-coord6-groups = 11 12
pull-coord7-groups = 11 12

pull-coord1-geometry = distance ; How is the coordinate defined? Here by the COM distance.
pull-coord1-type = external-potential ; Apply the bias using an external module.
pull-coord1-potential-provider = AWH ; The external module is called AWH!

awh = yes ; AWH on.
awh-nstout = 50000 ; Step interval for writing awh*.xvg files.
awh-nbias = 1 ; One bias, could have multiple.
awh1-ndim = 1 ; Dimensionality of the RC, each dimension per pull coordinate. pull-coord1-groups
awh1-dim1-coord-index = 1 ; Map RC dimension to pull coordinate index (here 1–>1)
awh1-dim1-start = 0.25 ; Sampling interval min value (nm)
awh1-dim1-end = 0.70 ; Sampling interval max value (nm)
awh1-dim1-force-constant = 128000 ; Force constant of the harmonic potential (kJ/(mol*nm^2))
awh1-dim1-diffusion = 5e-5 ; Estimate of the diffusion (nm^2/ps),used to initial update size, how quezly the system moves
awh1-error-init = 5 ; Estimate of the error of diffusion , used to set initial update size
awh-share-multisim = yes ; Share bias across simulations
awh1-share-group = 1 ; Non-zero share group index

ERROR 12 [file AWH3.mdp]:
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 12 is larger than
that and does not have a specific atom selected as reference atom.

Pull group natoms pbc atom distance at start reference at t=0
1 58792 29396
2 29706 29344 0.483 nm 0.000 nm
3 3720 29119
4 11160 29126 0.000 nm 0.000 nm
5 14890 29127
6 18432 29149 0.002 nm 0.000 nm
7 18457 29149
8 40335 29500 0.495 nm 0.000 nm
9 14816 29530
10 58792 29396 0.404 nm 0.000 nm
11 1914 59753
12 2028 61724 3.807 nm 0.000 nm
11 1914 59753
12 2028 61724 3.807 nm 0.000 nm

Any suggestion or advice would be very useful

Thank you so much!


Your problem is not with AWH, but with your setup with the pull code. You either have a very large pull group, or your pull group is split over periodic boundary conditions. In the first case you need to choose a pbcatom that is in the middle up the group. In the latter case, you need to put all atoms of the whole group in the same periodic image.

Got it !! Thank you so much
Since I have 10 chains, so I have to create 10 pull groups and define pbcatom that is in the middle of each chain. Am I right?

Thank you so much again, Dr.Hess

I’m very glad to send to you, dear hess. Recently, I read your tutorial on using the AWH method, which is used to research DNA base pair opening. This tutorial was written by
Viveca Lindahl. At the same times, I’m reading all articles about this method.The following:

However, There are some problems about the .mdp of AWH method.
I just would like to study the changes of free energy during unfolding of nonclassical DNA by AWH method, such as triplex DNA .

Question 1 :
How can I choose the Reaction coordinates? It’s same as the method of umbrella simple.
Question 2 :
Do I need to fix one end of the reaction coordinate during sampling?
Question 3:
How much should I set my force constant of the harmonic potential?
What is the difference of awh-multi and awh-simple?

Looking forward to your reply, thanks .

The manual and mdp options user guide explains most of these things.

You seem to be referring to an old tutorial.

The force constant should be set higher than the curvature of the free-energy landscape you are trying to sample.

Firstly, I’m very glad to your reply. I have read the AWH method of gmx manual. you say,“You seem to be referring to an old tutorial.”. Could you please tell me the new tutorial of the AWH method apply? Thank you very much, best regards.