On Tue, Dec 24, 2013 at 10:55 AM, houcemeddine <hothman.pasteur.fr> wrote:
> Hi everybody,
> I run some tests for MMPBSA.py to estimate the free energy using GB, and
> Nmode for a protein-protein complex (nearly 5500 atom) in order to rank
> some docking poses.
> Using maxcyc=10 and 20 runs the calculation simultaniously for 15 and
> 30 min for only one snapshot.
>
maxcyc=10 or 20 is WAY too low unless the trajectory has already been
minimized. It typically takes thousands of minimization steps for the
structures to minimize within allowable tolerances for the normal mode
calculation. When you set maxcyc=10 or 20, the structure will minimize,
realize it is not minimized enough, then quit without calculating the
normal modes (or at least that's what it should do). You should keep
maxcyc large, and the minimization will stop once the gradient falls below
a certain threshold (drms).
> I am aware that the entropy term is
> computationaly expensive, however in my case it seems that the
> calculation converges very slowly while my estimations of computation
> time is 10 days/snapshot
>
This is not unusual, and is the primary reason normal mode calculations are
often avoided when possible. The minimization procedure can take a very
long time, as can the hessian construction (and diagonlization) itself.
For very large systems, 10 days may be fairly typical.
Does someone have any idea about this issue?
>
> I also desired to use an input file to tune the parameters for Nmode
> with -use-mdins flag. The issue is that only the _MMPBSA_gb.mdin is
> generated at the end of the calculation but not
> _MMPBSA_sander_nm_min.mdin neither _MMPBSA_nmode.in were present even if
> I use the -make-mdins flag. is there any explanation for this ?
>
Many calculations in MMPBSA.py are now done with pipes and command-line
input rather than with temporary input files. Normal mode calculations are
done by passing the necessary parameters to mmpbsa_py_nabnmode via the
command-line, so no input file is ever created. This is not really much of
a limitation, though, as all of the available options are accessible
through the MMPBSA.py input file (unlike several options for GB and PB
calculations).
HTH,
Jason
--
Jason M. Swails
BioMaPS,
Rutgers University
Postdoctoral Researcher
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Tue Dec 24 2013 - 11:00:02 PST
