[gmx-users] Setting up a random system

[Jay Mashl]

On Sun, 9 Jul 2006, Mark Abraham wrote:
>> Hello,
>> I am trying to reproduce the assembly of phospholipids into a micel
>> and so I fill a box using the genbox command. And then I solvate it
>> using the same command.
>> I have tried really hard to have it running but I get a segmentation
>> fault at worst and at best a pressure scalling warning after a couple
>> steps and then a segmentation fault.
>> I use the -vdwd option so that the molecules aren't too close. If I
>> fill the box by using the lipid as a solvant then I can get a box to
>> work (although all the lipids are parallel to each other) but if I
>> make the box twice as big it crashes once more... I should add that I
>> manage to make it work fine with just the solvant, the problem occurs
>> when I put long chains.
>>
>> Would there be a solution to limit artificially the velocities of the
>> particles during the stabilization at the begining? Are there rules to
>> follow when setting up a system? It seems that it is a problem that
>> other people had but there is no clear methodology, I would be
>> greatfull to have some general tips or links on how to start a random
>> mix of solvant and lipids (or long molecules).
>> Thanks,
>> Erwin
>
> Have you minimised the energy to relieve bad contacts before starting MD?
> Have you tried position-restrained MD to start the equilibration process?
> Have you tried raising the temperature gradually rather than in one step?
>
> Mark

I find that solvating well-separated lipids can cause overlaps despite setting 
the -vdwd option sufficiently large.  The offending waters can be removed from 
the coordinate file by hand.

Jay