Re: [AMBER] MMGBSA doesn't finish

From: Jason Swails <>
Date: Fri, 24 Oct 2014 07:53:10 -0400

On Fri, Oct 24, 2014 at 2:22 AM, Bernhard Poll <>

> Hi everybody,
> I've got some issue with MMGBSA-calculation with Amber11. I'm
> calculating the energies for a larger dataset which works fine for a
> bunch of ligands but and some but at some point (which seems to be
> pretty random) the calculation doesn't finish which means it does write
> the all temporary files (_MMGBSA_gb_ligand.mdout etc. ) and at the end
> of each file it says "MM/GBSA processing done" but doesn't write the
> Results.dat-file. Interestingly if I re-ran the calculation it finished
> so there's no issue with the inputfiles I guess. I also double-checked
> the input files and the md-simulations by visualisation, acdoctor etc.
> and everything looks fine too. Maybe its some installation-issue but
> then it won't work at all?!We switched also from openmpi to mpd for
> parallelisation maybe that's an issue? Maybe some of you have an idea.

​This doesn't sound like an Amber issue. It sounds like an issue with your
system. Switching MPIs is OK, as long as you recompile Amber completely
with the new MPI. You can't all of a sudden start using MPICH2 wrappers
(like mpirun/mpiexec) when you used OpenMPI to compile Amber. It will not
work, and the behavior is unpredictable.

It could also be a filesystem issue, or if this is a cluster with a
submission system, the files may be getting "stuck" on a particular node
and not properly copied from the scratch disk to the NFS disk you are
expecting the files to be on.

​This will likely require some investigation on your part to see what is
wrong, but is not likely to be the culprit here.

One more comment -- Amber 11 (and AmberTools 1.5, presumably) are fairly
outdated and no longer supported (I would have to look around to even find
the source code anymore). If you encounter any issues that _do_ appear to
stem from Amber itself, our first question to you will likely be "does it
work with AmberTools 14?"

Good luck,

Jason M. Swails
Rutgers University
Postdoctoral Researcher
AMBER mailing list
Received on Fri Oct 24 2014 - 05:00:03 PDT
Custom Search