[gmx-users] Re: Pull code/starting structures

[Justin MacCallum]

Hi Cristian,

Unfortunately, the current state of the pull code is a bit of a mess. We
currently have an in house version here in Calgary that is somewhat
changed from the code in Gromacs 3, but its not even close to being
considered release quality.

The starting structure option that you are using is simply for generating
starting structures to use with the umbrella sampling or constraint forces
code.  The code currently works by adding a constant velocity
(translation_rate) in the correct direction.  The r0_group2 option only
applies if you have two pulled groups, which you don't seem to.  So the
setting of r0_group2 should have no effect on your system.  I'm not sure
why the direction options aren't working.

But, as I said the starting structure options are just for generating
starting structures.  In order to actually calculate the potential of mean
force, you need to use the umbrella sampling or constraint force options.
As far as I know, the umbrella sampling stuff that is in the 3.0.x release
is completely broken.  Even the version that I have is probably not that
suitable for the problem you are looking at.  Eventually, I will probably
clean up the pull code, add some extra features, and simplify the
interface.  In short, until the pull code is improved, I don't think that
it is that suitable for your problem.

Cheers,
Justin


> Hi everyone,
> 
> for my diploma thesis I have to simulate a protein with a small buried 
> ligand. I want to force the ligand to diffuse out of the protein, 
> taking a yet unknown path. Then I want to do some free energy 
> calculation with the mean force necessary for pushing/pulling the 
> ligand out.
> At first glance the starting structure pull code of Gromacs 3.0.5 
> seems to be the way to go (any other suggestions?), but I'm a little 
> bit confused about the whole algorithm in general and the parameters 
> in the .ppa input file in particular:
> 
> What exactly is the use of "r0_group2" and what shall I do with the 
> generated out_<something>.gro structures, which are created in the 
> given distance of the reference structure?
> I managed as a test to pull in vacuo one of the ligand molecules from 
> a distance of 2 nm to a (freezed) other one, but trying to push the 
> molecules away from another (with the reference ligand freezed again) 
> doesn't work, i. e. the parameter "reverse" seems always to be set to 
> "to_reference".
> In all cases I set the "r0_group2" to 0.0 0.0 0.0 (no 
> out_<something>.gro stuff was created then ;-)) and the "transstep" 
> and "tolerance" to zero.
> 
> And finally where do I get the force for pulling from (the .pdo file 
> seems to contain only coordinates)?
> 
> Any hints are much appreciated!
> 
> Thanks in advance for your help, Christian
> 
> PS: The freezing of the reference structure isn't the cause of my 
> "from_reference" problem, I just tried it without.