[gmx-users] use g_mindist to generate restraints

[Justin A. Lemkul]

Weingarth, M.H. wrote:
> 
> Dear gromacs users,
> 
> I wonder if it would be possible to generate restraints in the following 
> manner:
> 
> I do simulations of a peptide embedded in a lipid bilayer and I have 
> experimental NMR data which reflect the number of water molecules within 
> a radius of ~ 0.9 - 1.0 nm for several carbons of my peptide.
> 
> I could use these data to preorient my peptide to the membrane. I did a 
> run of ~ 100 ns and back-calculated the experimental data
> with g_mindist with the -on and -d 0.9 flags. This gives me the number 
> of water molecules around a certain carbon within a radius of 0.9 nm and 
> the MD matches the NMR data rather well actually.
> 
> My question would be if it would be possible to exploit these data as 
> restraints :
> 
> for instance :
> 
> Carbon C1 has ~ 100 - 300 water molecules within a radius of 0.9 nm
> Carbon C5 has ~ 50  - 150 water .............................0.9 nm
> Carbon C8 ......10  - 50 water ..............................0.9 nm
> 
> Once this would not be given, I woud like to apply a penalty function. 
> In this manner, I might be possible to place a peptide on the bilayer 
> surface and to use the number of sourrunding H20 as a driving force to 
> insert it in in the membrane without any preorientation.
> 
> Unfortunaly, I do not know how to generate such restraints - to create 
> .itp files with g_sas is not an option for what I calculate is very 
> different from the solvant accessible surface.
> 
> Thank a lot for any help & suggestions.
> 

This can be done, but it might be a nightmare.  You can't abstractly specify 
that a certain number (or range) of water molecules be within a certain distance 
of a given atom.  You have to specific a discrete number of water molecules, and 
exactly which water molecules those are.  To do this requires a merged 
moleculetype of protein + those waters in order to apply distance restraints. 
Then you can apply the penalty function as given in manual section 4.3.4.  These 
merged waters also then have to be consecutive with the protein in the 
coordinate file, or else some randomly placed, distant water molecule will have 
a strong distance restraint placed on it at step 0 and the whole thing may 
collapse.

Note that many simultaneous distance restraints will limit (or completely 
obliterate) your ability to use domain decomposition.  You'll almost certainly 
have to run with mdrun -pd or in serial.  You will also have to compensate for 
the fact that you cannot have multiple [settles] directives for rigid water 
(i.e. for the restrained ones and the rest of the bulk water).  You'd have to 
convert all of SETTLE constraints of the merged the water molecules to normal 
constraints (three for each molecule, O-H1, O-H2, H2-H2), which may have 
stability problems.

-Justin

> Cheers
> 
> Markus
> 

-- 
========================================

Justin A. Lemkul
Ph.D. Candidate
ICTAS Doctoral Scholar
MILES-IGERT Trainee
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================