PBS script for BEAST v1.8.0

167 views
Skip to first unread message

pawan jayaswal

unread,
Mar 9, 2016, 5:33:50 AM3/9/16
to beast-users
Hello Beast User,

  I have just try to run BEAST v1.8.0 on HPC cluster from qsub:

#!/bin/sh
### Job name
#PBS -N beast1kmn
### Output files
#PBS -o pbs.log
#PBS -l walltime=480:00:00
java -Xmx4048m -Djava.library.path=/opt/software/beagle/lib -jar /opt/software/BEAST/BEASTv1.8.0/lib/beast.jar -beagle_CPU Inputfile.xml >log.txt

but it wont run and neither showed any error.

Any correction or suggestion will be highly appreciated.

JBagz

unread,
Mar 9, 2016, 7:54:49 AM3/9/16
to beast-users
Hi Pawan,

Try the following suggestions. First, change the first line of code to say bash instead of sh, as bash or csh shells are most common in my experience; so this could be the issue. Second, you are also missing pertinent information after your resource info flag (-l), like the number of nodes. Third, your wall time is incredibly excessive. Do you realize that 480 hours is 20 full days (!) and yet most BEAST runs should finish in 7 or less? This may exceed allocations for your cluster, which often have walltime limits. (Of course, your limit may be 480 h, but why the overkill?) Other issues are that you probably need to specify the working directory, and we do this on a cluster by adding "cd $PBS_O_WORKDIR" followed by "exit 0" to the end of the code; this is just optional, for you to try. The other way to do this is to "cd" to the working directory before the beast commands ("java..."), assuming the input file is in the working directory. Last, I don't think you need the pbs output log, as you should get standard error and out anyway, and I would also change the beast output file (after ">") to have the extension ".out". 

Here is an edited version of the file, which I would try, but remember to name with extension ".sh":

#!/bin/bash
#---Job name 

#PBS -l nodes=1:ppn=1,pmem=2gb,walltime=168:00:00
#PBS -N beast1kmn

java -Xmx4048M -Djava.library.path=/opt/software/beagle/lib -jar /opt/software/BEAST/BEASTv1.8.0/lib/beast.jar -beagle_CPU Inputfile_name.xml > Inputfile_name.out

#--working directory
#--option1:
cd $PBS_O_WORKDIR
exit 0 


Please let me know if this runs for you. Thanks.

Best,

Justin

Justin C. Bagley, Ph.D.
CNPq Young Talent Fellow (Postdoc) and Research Associate
Coleção Herpetológica
Departamento de Zoologia
Universidade de Brasília
Campus Universitário Darcy Ribeiro
70910-900 Brasília, DF, Brasil
jcbagley _at_ unb.br
Website: http://www.justinbagley.org

JBagz

unread,
Mar 9, 2016, 8:02:33 AM3/9/16
to beast-users
Just to be clear, you should submit the job to the PBS queue by using the command qsub input_file.sh or qsub input_file.pbs, and also be aware that cd $PBS_O_WORKDIR tells the system to make the run from the same directory that you executed the qsub command from.

Best,

Justin
Reply all
Reply to author
Forward
0 new messages