PrepareJob

From BIOcrunch
Revision as of 15:45, 4 July 2015 by Sschmidt (Talk | contribs)

Jump to: navigation, search

To run a job on the cluster, prepare the data files on your local machine, then copy it to the cluster. Beside the data files, you will need a job file with the settings as outlined below.

Name your job file and your data files in a consistent way:

  • Job file: myjobfile.sh
  • User: Mustermann
  • Email: max@mustermann.de

The job files are written in a sort of "shell" command syntax that is common on Linux systems.

Lines starting with "##" or "# " are comment lines to document some stuff for better reading afterwards.

Lines starting with "#$" are commands for the job scheduler "qsub".

Lines starting not with "#" are commands to execute programs in this job.

Contents

Examples

You may use one of the following examples as a template for you job. Lines that need to be modified are highlighted. Make sure to change username, email adress, filenames, and, if necessary, program options. The last line of each script calls the program you want to run, including all parameters. The number of CPUs is stored in $NSLOTS and is set later when starting the job.

Beast

#!/bin/csh
#
## Username (Which account to be charged cpu time)
#$ -A Mustermann
#
## Send mail to users
#$ -M max@mustermann.de
#
## Mail at beginning/end/on suspension
#$ -m bes
#
## Export these environmental variables
#$ -v PVM_ROOT,LD_LIBRARY_PATH=/share/apps/beaglenew
#
## The job is located in the current working directory.
#$ -cwd
## Filenames for output and error log files
#$ -o myjobfile.out
#$ -e myjobfile.err
#
env >./xenv
/share/apps/beast180/bin/beast -beagle -beagle_CPU -beagle_instances $NSLOTS -overwrite ./myjobfile.xml > ./myjobfile.out

Beast2

#!/bin/csh
#
## Username (Which account to be charged cpu time)
#$ -A Mustermann
#
## Send mail to users
#$ -M max@mustermann.de
#
## Mail at beginning/end/on suspension
#$ -m bes
#
## Export these environmental variables
#$ -v PVM_ROOT,LD_LIBRARY_PATH=/share/apps/beaglenew
#
## The job is located in the current working directory.
#$ -cwd
## Filenames for output and error log files
#$ -o myjobfile.out
#$ -e myjobfile.err
#
env >./xenv
/share/apps/beast210/bin/beast -beagle -beagle_CPU -beagle_instances $NSLOTS -overwrite ./myjobfile.xml > ./myjobfile.out

Mr Bayes

#!/bin/csh
#
## Username (Which account to be charged cpu time)
#$ -A Mustermann
#
## Send mail to users
#$ -M max@mustermann.de
#
## Mail at beginning/end/on suspension
#$ -m bes
#
## Export these environmental variables
#$ -v PVM_ROOT,LD_LIBRARY_PATH=/share/apps/beaglenew
#
## The job is located in the current working directory
#$ -cwd
## Filenames for output and error log files
#$ -o myjobfile.out
#$ -e myjobfile.err
#
env >./xenv
mpirun -np $NSLOTS /share/apps/mrbayes/mb322 ./beispiel.nex </dev/null

Raxml

#!/bin/csh
#
## Username (Which account to be charged cpu time)
#$ -A Mustermann
#
## Send mail to users
#$ -M max@mustermann.de
#
## Mail at beginning/end/on suspension
#$ -m bes
#
## Export these environmental variables
#$ -v PVM_ROOT
#
## The job is located in the current working directory.
#$ -cwd
## Filenames for output and error log files
#$ -o myjobfile.out
#$ -e myjobfile.err
#
env >./xenv
mpirun -np $NSLOTS /share/apps/raxml/raxmlHPC-MPI-SSE3.icc -s sequencefile.phy -n outputfile.phy -m PROTGAMMAWAG
Personal tools
Namespaces

Variants
Actions
Navigation
Tools