qsub_clark.sh

#!/bin/bash                                                                                             
#                                                                                                       
#$ -j y                         
# join std out & err                                                    
#$ -cwd                         
# run job and store logs in cwd                                         
#$ -m ea                      
# send mail on begin, end, & abort                                        
#$ -M [email protected]  # email address                                                         
#$ -N clark_tooloff             # job name                                                              
#$ -l h_rt=99:00:00      # wall time request of 5 hour                                                  
#$ -pe smp 1             # parallel environment, 18 cores                                               
#$ -l vf=40G              # memory needed 5G per core                                                   
#$ -l h_vmem=60G          # max mem request of 10 GB per core                                           
#$ -l os=rhel6.3         # use cores running rhel6.3                                                    
#$ -V                    # import environment                                                                                                                                                                                                                                                                           

REMOTE_IN=$1                                                                                            
IN=${TMPDIR}/`basename $REMOTE_IN`                                                                      
CLASSIFY_TIME=/home/dcd3001/tool-off/clark/results/`basename ${IN}.clark_classify.time`                 
ABUND_OUT=${IN}.abund.csv                                                                               
ABUND_TIME=/home/dcd3001/tool-off/clark/results/`basename ${IN}.clark_estimate.time`                                                                                                                            

rsync -av $REMOTE_IN $IN                                                                                                                                                                                        
cd /home/dcd3001/tool-off/clark/CLARKSCV1.2.3/                                                                                                                                                                  

echo 'classify_metagenome'                                                                              
/usr/bin/time -f "%e wall\t%U user\t%M maxmem" ./classify_metagenome.sh -O $IN -R $IN --gzipped 2> $CLASSIFY_TIME                                                                                               
rsync -av ${IN}.csv /home/dcd3001/tool-off/clark/results/`basename $IN`.csv                                                                                                                                     

echo 'estimate abundances'                                                                              
/usr/bin/time -f "%e wall\t%U user\t%M maxmem" ./estimate_abundance.sh -D /home/dcd3001/tool-off/clark/DB_DIR -F $CLASSIFY_OUT > $ABUND_OUT 2> $ABUND_TIME                                                      
rsync -av $ABUND_OUT /home/dcd3001/tool-off/clark/results/`basename $ABUND_OUT`