[gmx-users] Inconsistent shifts using multiple bonds type 6

[Mark Abraham]

> Hello all,
>
> I am trying to apply a set of harmonic distance restraints between pairs
> of
> protein atoms in my system, which is protein/ligand/water/ions in a
> periodic box.
>
> The way i am doing this is to use the bond type 6 and to add these bonds
> at
> the end of the [bonds] section in my topology file, after the system has
> been energy minimised and i have performed position restrained MD, without
> these harmonic distance restraints.
>
> I can then run it through grompp with no problems and start mdrun. If i
> only have a few (ie less than about 30) harmonic distance restraints then
> the simulation if fine and runs happily. However if i increase the number
> of restraints then mdrun crashes due to lincs warnings and it also says
> that the are inconsistent shifts, (the number of inconsistent shifts
> varies
> depending on which atoms i choose to restrain and also the number of atoms
> chosen) a warning not seen with less harmonic distance restraints.

Why do you need more harmonic restraints?

> I should also say that i have run this system without these harmonic
> restraints without problems for tens of nanoseconds already. Also this
> problem can be repeated if i try to use the [distance_restraints] however
> it requires less restraints to get the inconsistent shift warning (and
> lincs problems) this way.
>
> If anyone can give me an idea of what is causing this problem it would be
> warmly appreciated, or even just how to detect which of the harmonic
> distance restraints will cause the problem before i run mdrun so i can
> replace the problem 'bonds' in the topology file.

You can't tell a priori, but if you watch the trajectory you should start
to get an idea what is breaking where and this will narrow down the number
of inappropriate restraints.

Mark