[gmx-users] Lincs constraints failure

[Mark Abraham]

Hi,

If the configuration and the Hamiltonian are too distant from each other,
then there can be no way to satisfy constraints. This happens frequently
during equilibration, but Hamiltonian exchange is another time it is
possible to generate a problem, if the new context is too far from the old.
That you haven't yet observed it with smaller (ie more forgiving) time
steps suggests this is the issue. I would look carefully at the history
leading up to the crash for clues. I would also consider running more
replicas, with intermediate values of the scaling parameter.

Mark

On Sun, Jun 25, 2017 at 11:09 PM Qinghua Liao <scorpio.liao at gmail.com>
wrote:

> Hello Mark,
>
> Thanks a lot for your reply! I performed Hamiltonian replica exchange
> simulations,
> the Hamiltonians (LJ, electrostatic, dihedrals ) of certain residues
> were scaled for exchange  between replicas.
>
> You mean this kind of constraints failure is quire common in replica
> exchange? Apart from using smaller time step (i.e. 1 fs),
> is there some other solution? Thanks a lot!
>
>
> All the best,
> Qinghua
>
> On 06/25/2017 09:04 PM, Mark Abraham wrote:
> > Hi,
> >
> > What are you exchanging between replicas? Seems like the most likely
> issue.
> >
> > Mark
> >
> > On Sat, 24 Jun 2017 16:50 Qinghua Liao <scorpio.liao at gmail.com> wrote:
> >
> >> Hello,
> >>
> >> I have searched on Google for the LINCS warning problems, and there are
> >> already many discussions,
> >> but I guess that my situation might be different.
> >>
> >> I have a system of protein/micelles/water_ions in a dodecahedral water
> >> box. The system is fully equilibrated by several steps.
> >>
> >> 1. minimizations (steep, cg) with positional restraints (1000 kJ/mol
> >> nm2) on protein and micelles,
> >> 2. 500 ps NVT equilibration, time step 1 fs, positional restraints (1000
> >> kJ/mol nm2) on protein and micelles,
> >> 3. 500 ps NPT equilibration, time step 1 fs, positional restraints (1000
> >> kJ/mol nm2) on protein and micelles,
> >> 4. 500 ps NPT equilibration, time step 1 fs, positional restraints (800
> >> kJ/mol nm2) on protein and micelles,
> >> ...500 ps, restraints, 600 kJ/mol nm2, 1fs
> >> ...500 ps,  restraints, 400 kJ/mol nm2, 1fs
> >> ...500 ps restraints, 200 kJ/mol nm2,  1fs
> >> ... 500 ps restraints, 100 kJ/mol nm2, 1 fs
> >>
> >> 5. 15 ns equilibration, positional restraints (100 kJ/mol nm2) only on
> >> protein, letting the micelles assemble, time step 1 fs
> >> 6. 5 ns MD simulation, without any restraints, time step 2 fs
> >>
> >> For all these steps, the constraints were applied to h-bonds, which is
> >> the same for following simulations.
> >> And the I started a Hamiltonian replica exchange simulation, 4 replicas
> >> were chosen, time step is 2fs.
> >> But the simulation crashed at the simulation time of 25 ns. The error is
> >> shown like:
> >> #
> >> step 12855987: Water molecule starting at atom 66929 can not be settled.
> >> Check for bad contacts and/or reduce the timestep if appropriate.
> >> Wrote pdb files with previous and current coordinates
> >> Wrote pdb files with previous and current coordinates
> >> #
> >> Step 12855991, time 25712 (ps)  LINCS WARNING in simulation 3
> >> relative constraint deviation after LINCS:
> >> rms 0.000678, max 0.010615 (between atoms 3541 and 3542)
> >> bonds that rotated more than 30 degrees:
> >>    atom 1 atom 2  angle  previous, current, constraint length
> >>      3541   3543   90.0    0.1010   0.1020      0.1010
> >> #
> >>
> >> Atom 3541 is ND2 of ASN227, and atom 3543 is HD22 bonded to atom 3541. I
> >> checked the trajectory file, and I didn't find Asn227 doing something
> >> improperly.
> >>
> >> I also found that the constraints failure happened faster if I used 2 fs
> >> with constraints for all bonds. I believe that a time step of 1 fs will
> >> work, which also means
> >> more computing time since the system is very big. Any suggestions?
> >> Thanks very much!
> >>
> >>
> >>
> >> All the best,
> >> Qinghua
> >>
> >> --
> >> Gromacs Users mailing list
> >>
> >> * Please search the archive at
> >> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> >> posting!
> >>
> >> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
> >>
> >> * For (un)subscribe requests visit
> >> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> >> send a mail to gmx-users-request at gromacs.org.
> >>
>
> --
> Gromacs Users mailing list
>
> * Please search the archive at
> http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before
> posting!
>
> * Can't post? Read http://www.gromacs.org/Support/Mailing_Lists
>
> * For (un)subscribe requests visit
> https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or
> send a mail to gmx-users-request at gromacs.org.
>