AMBER Archive (2006)

Subject: Re: AMBER: How to run mm_pbsa_statistics.pl?

From: Peng Tao (amberhelp_at_gmail.com)
Date: Thu Jan 05 2006 - 07:39:19 CST

Dr. Holger Gohlke,

Thank you very much for your replying. I tried your suggestion. It worked
for me, because my job run for about 6 hours. But I have difficulty to deal
with the memory. Since I have three intermediate files (file size in
parenthesis):
fkbp12PrbPep_com.all.out (1533601901byte)
fkbp12PrbPep_lig.all.out (18000401byte)
fkbp12PrbPep_rec.all.out (1219320401byte)

I don't know how much memory I need to assign for MM_PBSA batch job for this
calculation. In my last try, I assign 500MW, which are about 4000MB. It is
more than the total size of three files, which are 2771MB. However, my batch
job aborted after 6 hours run.

Here are the information I got from the PBS system:
Job Name: MMPBSAStatis
Execution terminated
Exit_status=0
resources_used.cput=06:42:41
resources_used.mem=3984128kb
resources_used.vmem=3984128kb
resources_used.walltime=06:43:26

Total memory the system used is 3984128kb, which is very closed to the upper
limit I assigned to this job.

The PBS log file of this job read:
=>> Values of global variables
    TEMP = 300
    R = 8.314
    gammaP = 0.00542
    betaP = 0.92
    gammaG = 0.0072
    betaG = 0
=>> Reading input
=>> Reordering files
    Final order:
    1. fkbp12PrbPep_com.all.out: 1 - 120
    2. fkbp12PrbPep_rec.all.out: 1 - 107
    3. fkbp12PrbPep_lig.all.out: 108 - 120
=>> Reading files
    Reading fkbp12PrbPep_com.all.out
Killed

The PBS job files I used:

#PBS -N fkbpPrbDKMMPBSA11Statis
#PBS -l walltime=92000
#PBS -l mem=500mw
#PBS -l nodes=1:ppn=1
#PBS -j oe
#PBS -m ae
#PBS -S /bin/csh

cd /home/workdirectory

mm_pbsa_statistics.pl 1 2 fkbp12PrbPep_statistics.in
fkbp12PrbPep_statistics.out
The job has been killed when it was still reading fkbp12PrbPep_com.all.out,
or right after finish reading it. From the output I found nothing wrong. But
the job didn't go through successfully. My only suspect is short of memory.
But I am not sure about this. Even it is true, I don't know how much memory
I actually need to assign for this job.

Could you please help me out here?

Thank you very very much.

Best regards,
Peng

On 1/4/06, Holger Gohlke <gohlke_at_bioinformatik.uni-frankfurt.de> wrote:
>
> Am Mittwoch, 4. Januar 2006 14:51 schrieb Peng Tao:
> > Dear all,
> >
> > I am trying to use MM_PBSA to decompose interactions in a
> protein-peptide
> > complex to residue pairs. For some reason, my calculation aborted
> > unexpectedly. Probably the sizes of intermediate output files are too
> > large.
> >
> > But I do have three intermediate files for final calculation:
> > mysystem_com.all.out(1.5Gb) mysystem_lig.all.out(18Mb) and
> > mysystem_rec.all.out(1.2Gb).
> > I am thinking about running script mm_pbsa_statistics.pl to read these
> > three files and generate final output, but don't know how.
> > The parameter file (mysystem_statistics.in) for mm_pbsa_statistics.pl is
> > very simple. There are only three lines:
> > mysystem_com.all.out 1_120
> > mysystem_rec.all.out 1_107
> > mysystem_lig.all.out 108_120
> > (Am I right about this?)
> >
> > But trying to run script mm_pbsa_statistics.pl by typing
> > mm_pbsa_statistics.pl < mysystem_statistics.in
> > simply didn't work.
> > What I got is:
> > USAGE: mm_pbsa_statistics.pl <calc delta ? 0..2> <calc decomp ? 0..2>
> > <input file> <output file>
> > [<snap_min> <snap_max>]
>
> If you want to run mm_pbsa_statistics.pl for a single traj. version of
> MM-PBSA, you need to set calc_delta to "1", for a three traj. version to
> "2".
> If you want to calculate a decomposition per residue, you need to set
> calc_decomp to "1", for a pairwise decomposition to "2", for no decomp. to
> "0".
>
> So the call for a 3-traj. version with decomp. per residue would be:
>
> mm_pbsa_statistics.pl 2 1 mysystem_statistics.in mysystem_statistics.out
>
> With snap_min, snap_max you can optionally restrict the number of
> snapshots to
> consider in the analysis to the intervall of [snap_min, snap_max].
>
> Let me know if it doesn't work!
>
> Best regards
>
> Holger
>
> >
> > Based on this information, I coulnd't figure out how to run
> > mm_pbsa_statistics.pl correctly.
> >
> > Could you please help me about this?
> >
> > Many thanks in advance.
> >
> > My MM_PBSA input file:
> >
> ###########################################################################
> >##### @GENERAL
> > PREFIX fkbp12PrbPep
> > PATH ../01_GenerateSnapshots/
> > #
> > COMPLEX 1
> > RECEPTOR 1
> > LIGAND 1
> > #
> > COMPT ../fkbp12PepCom.top
> > RECPT ../fkbp12.top
> > LIGPT ../probe_peptide.top
> > #
> > GC 0
> > AS 0
> > DC 1
> > #
> > MM 1
> > GB 1
> > PB 0
> > MS 0
> > #
> > NM 0
> > #
> >
> ###########################################################################
> >##### @DECOMP
> > DCTYPE 4
> > #
> > COMREC 1-107
> > COMLIG 108-120
> > COMPRI 1-120
> > RECRES 1-107
> > RECPRI 1-107
> > RECMAP 1-107
> > LIGRES 1-13
> > LIGPRI 1-13
> > LIGMAP 108-120
> > @MM
> > DIELC 1.0
> > @GB
> > IGB 2
> > GBSA 2
> > SALTCON 0.00
> > EXTDIEL 80.0
> > INTDIEL 1.0
> > #
> > SURFTEN 0.0072
> > SURFOFF 0.00
> >
> >
> >
> > --
> > Best regards,
> > Peng Tao
> >
> > Graduate Student
> > Chemistry Department
> > The Ohio State University
>
> --
> ++++++++++++++++++++++++++++++++++++++++++++++++++
> Dr. Holger Gohlke
> Juniorprofessur fuer Molekulare Bioinformatik
>
> J.W. Goethe-Universitaet
> Fachbereich Biowissenschaften
> Institut für Zellbiologie und Neurowissenschaft
> Marie-Curie-Str. 9
> 60439 Frankfurt/Main
> Germany
>
> Tel.: (+49) 69-798-29411; Fax: (+49) 69-798-29464
> Email: gohlke_at_bioinformatik.uni-frankfurt.de
> URL: http://www.uni-frankfurt.de/~hgohlke
> ++++++++++++++++++++++++++++++++++++++++++++++++++
> -----------------------------------------------------------------------
> The AMBER Mail Reflector
> To post, send mail to amber_at_scripps.edu
> To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu
> 

--
Best regards,
Peng Tao

Graduate Student
Chemistry Department
The Ohio State University

-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu