[gmx-users] Targeted Molecular dynamics

[Mark Abraham]

Olivier Perin wrote:
> Dear gmx users,
> 
> OBJECT:
> I just contact you to confirm one thing about the ability to run a 
> targeted molecular dynamic simulation with gromacs with the help of 
> positional restraints.
> 
> SITUATION:
> If struct1 and struct2 are two near conformations of one same protein 
> could I use these two gromacs commands in order to force the first 
> system (struct1) to direct to the second system (struct2)?

It will take a finite number of MD steps to get arbitrarily close to 
struct2, but yes this should work.

> MY HYPOTHESIS:
> pdb2gmx -f struct1.pdb -i posre.itp -posrefc 1000 -p struct1.top
> 
> grompp -f MD_050K.mdp -c struct1.pdb -p struct1.top -o struct1_MD.tpr -r 
> struct2.pdb
> 
> (for me, the option -r of grompp allowed to define with which 
> coordinates the positional restraints must be set)
> 
> mdrun -s struct1_MD.tpr  -o struct1_MD.trr   -x struct1_MD.xtc  -c 
> struct1_DM1.pdb
> 
> QUESTIONS:
> Do these commands seems to be correct?

Yes

> What do you think about 1000 for positional restraints? Is it to high?

I suspect so, but if they are too high, then something in the structure 
will break fairly quickly. If it's a complex structural change, then 
1000 is almost certainly too high. Try it and see :-)

Mark