[gmx-users] Re: umbrella sampling for two polymer interaction

Thomas Schlesier schlesi at uni-mainz.de
Fri May 31 12:21:44 CEST 2013

Look also into the manual. But the tutorial is a nice place to start.
For further comments see below:

> Dear Lloyd,
> I have read that but my system is different
> regards,
> On Thu, May 30, 2013 at 8:28 PM, lloyd riggs<lloyd.riggs at gmx.ch>  wrote:
>> >Dear Jiom,
>> >
>> >Look at justines tutorial, there's example pull .mdp.
>> >
>> >Stephan Watkins
>> >

>> >
>> >I want to do Umbrella sampling between two different polymers (A and B)
>> >interacting with each other with starting configuration separated by some
>> >distance and I am trying to bring them closer.
>> >
>> >I have some queries regarding pull inputs: (this is for to run a umbrella
>> >sampling at some distance)
>> >
>> >pull = umbrella
>> >pull_geometry = distance
>> >pull_dim = Y Y Y
>> >pull_start = ???
>> >pull_ngroups = 2?
>> >pull_group0 = polymer_B
>> >pull_group1 = polymer_A
>> >pull_init1 = 0
>> >pull_rate1 = 0.0
>> >
>> >
>> >please suggest for following:
>> >
>> >1) pull_dim I have set to Y Y Y: Is this correct I do not want to make
>> >it interact with some directional vector

I don't really understand the question. If you use 'pull_dim = Y Y Y' 
the pulling potential acts in all 3 dimensions, if you use 'pull_dim = Y 
N N' the pulling potential acts only in the X direction and your two 
groups an move freely in the YZ-plane.

>> >
>> >2) Which should be group0 or group1, in other words should I pull both
>> >together or how I should decide which one should be reference and
>> >which to be pulled as both are different polymers?
Depends on what you want to do. Easiest way would be define one polymer 
a group0/reference group and the other as group1/pulled group. System 
shouldn't care about which polymer is which group.
If you do a pulling simulation, there can be reason for chosing the 
groups (protein = reference , ligand = pulled group, since we want to 
pull it away)

>> >
>> >3) And also what should be pull_ngroups because if there is no
>> >reference group then it should be 2
Better use a reference group -> pull_ngroups = 1
You don't want to pull in absolute coordinates, when your system can 

>> >
>> >4) I am not able to understand pull_start option with pull_init1. In
>> >this case if it is set to yes and 0.0 respectively then does that mean
>> >this combination is equivalent to pull_start = No if I just assume
>> >pull_init1 does not have any default value (which is 0.0); not
>> >existing
 From the setup which you have written above:
polymer_B is the reference group. the origin of the pulling potential is 
at 'pull_init1' (a length) along the vector which goes from polymer_B to 
polymer_A (sine you use pull_geometry = distance).
If you set pull_init1=0 and pull_start=no, polymer_A will crash into 
polymer_B (since the origin of the umbrella potential is directly at the 
center of mass of polymer_B).
If you set pull_init1=0 and pull_start=yes, GROMACS adds the distance 
between polymer_B and A to pull_init1 (-> so pull_init1 is now greater 
than 0.0). Now the origin of the umbrella potential is at the center of 
mass of polymer_A. -> A is restrainted to a certian distance of B.

>> >
>> >5) Also finally where are upper and lower bounds defined. pull_k1 =
>> >1000 is harmonic applied to some equilibrium distance value. How this
>> >distance is taken by the programme (or it is just the starting
>> >distance taken between two groups) and what are the ± values
>> >defined. (say in AMBER I define r1,r2,r3,r4; where r2=r3 which is
>> >assumed equilibrium value and r1 is lower and r4 is upper value which
>> >defines shape of potential)
The umbrella potential is a simple harmonic potential (so no fancy stuff 
as in AMBER) with
V = 1/2 k x^2
where x is the violation of the equilibrium distance.
For your setup
V = 1/2 (pull_init1 - distance(B-A))^2
where distance(B-A) means the distance between both polymers.


More information about the gromacs.org_gmx-users mailing list