I just noticed that bugfix #28 for Amber10 ³contains a few updates to NMR
restraints in sander and pmemd². Is it possible that additional problems
were caused by this patch?
The binary I have issues with was compiled using bugfixes 1-35.
The binary that works was complied using bugfixes 1-17.
Doug
On 5/21/10 2:30 PM, "Douglas Kojetin" <douglas.kojetin.gmail.com> wrote:
>
> In this case, Iım pretty sure my restraint energies should not be 0.
>
> I also performed a serial simulation (Amber 10, sander) and see similar
> results (Bond = 0).
>
> I think it could be something wrong with these specific Amber10
> sander/sander.MPI binaries, as calculations on a different computing cluster
> (Amber 10 binaries, but compiled a while ago) reports restraint energies > 0.
> We only have Amber 10 bug fixes 1-32 for the problematic binaries, and the
> current is 35. Bugfixes 1-17 were applied to the older binary that reports
> restraint energies > 0.
>
> I havenıt this specific dataset on the older binaries (yet), but I noticed
> this problem (Bond = 0) previously on another dataset, switched to another
> cluster running Amber 10 (an older complied version), and the calculations
> gave results with Bond > 0.
>
> I suppose we should recompile and hope for the best ...
>
> Thanks,
>
> Doug
>
>
>
>
> On 5/21/10 2:19 PM, "case" <case.biomaps.rutgers.edu> wrote:
>
>> On Fri, May 21, 2010, Douglas Kojetin wrote:
>>> >
>>> > I am trying to perform protein structure refinement using AMBER 10
>>> > (sander.MPI). My input restraint file (DISANG.RST) contains 888 NMR NOE
>>> > distance restraints and 336 torsion angle restraints (84 dihedral angles,
>>> > 252 chirality restraints). The total number of restraints = 1224.
>>> >
>>> > The full output file is attached
>>
>> Specific advice:
>> It is probably the case that all your restraint energies are zero. The
>> summary printed at the end only prints restraints whose energy is greater
>> than
>> pencut (I think that is the name), which defaults to some positive value.
>> Set pencut=-1.0 (or any negative number) to see all of the restraints in the
>> summary at the end.
>>
>> General advice:
>> Try a serial simulation, to eliminate the possibility of specific parallel
>> problems. Don't run a many-step simulation with a thousand restraints: set
>> up
>> a 10 step run (with ntpr=1) and use a single restraint. The command
>> "LISTIN=POUT" is also useful in seeing how sander is interpreting your input.
>>
>> ....dac
>>
>>
>> _______________________________________________
>> AMBER mailing list
>> AMBER.ambermd.org
>> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri May 21 2010 - 12:30:06 PDT
