[gmx-developers] User-defined potential

[Erik Lindahl]

mustafa toprakci wrote:

>  Thanks Erik,
>  What I would like to do is to prevent the atoms of a ligand to get 
> too close to the atoms of the receptor when I heat the system of these 
> two molecules at high temperatures (~1000 K).
>  I just want to add this restraint to the non-bonded interactions. I 
> hope this will prevent ligand bump to the receptor. I will be pleased 
> if you show me the way to add this in Gromacs and which table to 
> select for this half-harmonic term.

The repulsive part of the Lennard-Jones interaction already does this, 
so I guess your problem might rather be that the high temperature causes 
the simulation to crash when atoms collide?

Study the 'Force Field' and 'Special Topics' (the pull code) sections of 
the manual - here are some ideas:

1. You might be able to use position restraints on the ligand and/or 
receptor to avoid the atomic coordinates from changing too much.
2. You could use soft-core interactions to remove the high peak of the 
repulsive interaction - this will avoid crashes at high temperatures.
3. You could use the pull package and put a distance restraint between 
the entire ligand and receptor groups - this is probably the best 
solution. The pull code might not be very well documented, but Peter & 
Justin read this list, so I'm sure they can help with questions :-)

Cheers,

Erik