CheckMerkers.pbs

#!/bin/bash

#PBS -N missing 
#PBS -l nodes=1:ppn=1
#PBS -l walltime=60:00
#PBS -q hopper

# ===== ATOOLS SCRIPT =====
#
# READTHEDOCS : https://atools.readthedocs.io/en/latest/
#
# SUBMIT USING :
# $ module load atools/torque
# $ n=$(( $(wc -l samples.csv | cut -f1 -d' ') -1 )) 
# $ qsub -N merkers -t 1-$n CheckMerkers.pbs
#
# FOLLOW JOB RUN WITH :
# $ arange --sniff 1 --log <jobname>log<jobid> --summary
# $ arange -t1-$n --log <jobname>log<jobid> --summary
# ( does not work : $ aload --log <jobname>log<jobid> )
#
# POST PROCESSING :
# $ areduce -t 1-$n --pattern "merkers-{PBS_ARRAYID}.tmp" --out merkers.out 
# $ rm *.tmp
# $ rm merkers.[oe]<jobid>-*
#
# =========================


export PATH="/apps/antwerpen/broadwell/centos7/atools/1.4.7-torque/bin/../lib/../bin:${PATH}"

alog --state start

cd "$PBS_O_WORKDIR"

Mini1_F="TGAGGTCTGAGGTACTTCGAAAG|$(./complement.sh TGAGGTCTGAGGTACTTCGAAAG)"
Mini1_R="TGGATTACTTGGTGTTTTCTATTGATAA|$(./complement.sh TGGATTACTTGGTGTTTTCTATTGATAA)"
Mini1_P="TTTTCCTGGAGAAAACTGTAT|$(./complement.sh TTTTCCTGGAGAAAACTGTAT)"

Mini2_F="TCTTATGACTGATTTTACGAGAATA|$(./complement.sh TCTTATGACTGATTTTACGAGAATA)"
Mini2_R="GACATAACAGAGGAAAGTGCTC|$(./complement.sh GACATAACAGAGGAAAGTGCTC)"
Mini2_P="TTGTGGTAAGAGTGATTTAGTAAT|$(./complement.sh TTGTGGTAAGAGTGATTTAGTAAT)"

Mini3_F="AAACCAACAGAAAGAGATTGCTTA|$(./complement.sh AAACCAACAGAAAGAGATTGCTTA)"
Mini3_R="ATGGTGATAGAAGTTAGAGATGTGTAG|$(./complement.sh ATGGTGATAGAAGTTAGAGATGTGTAG)"
Mini3_P="TAGATGTAGTATAAGAATTTAAAAT|$(./complement.sh TAGATGTAGTATAAGAATTTAAAAT)"

TgsGP_F="GAAGCAGTGGGACCTTAGC|$(./complement.sh GAAGCAGTGGGACCTTAGC)"
TgsGP_R="TTTGTGCTCTTGCTTGCTATTAC|$(./complement.sh TTTGTGCTCTTGCTTGCTATTAC)"
TgsGP_P="CTCTCCGAACACAGCAGCGACATC|$(./complement.sh CTCTCCGAACACAGCAGCGACATC)"

q18S_F="CGTAGTTGAACTGTGGGCCACGT|$(./complement.sh CGTAGTTGAACTGTGGGCCACGT)"
q18S_R="ATGCATGACATGCGTGAAAGTGAG|$(./complement.sh ATGCATGACATGCGTGAAAGTGAG)"
q18S_P="TCGGACGTGTTTTGACCCACGC|$(./complement.sh TCGGACGTGTTTTGACCCACGC)"

GPI_PLC_F="CCCACAACCGTCTCTTTAACC|$(./complement.sh CCCACAACCGTCTCTTTAACC)"
GPI_PLC_R="GGAGTCGTGCATAAGGGTATTC|$(./complement.sh GGAGTCGTGCATAAGGGTATTC)"
GPI_PLC_P="ACACCACTTTGTAACCTCTGGCAGT|$(./complement.sh ACACCACTTTGTAACCTCTGGCAGT)"


all_probes=($Mini1_F $Mini1_R $Mini1_P \
		$Mini2_F $Mini2_R $Mini2_P \
		$Mini3_F $Mini3_R $Mini3_P \
		$TgsGP_F $TgsGP_R $TgsGP_P \
		$q18S_F $q18S_R $q18S_P \
		$GPI_PLC_F $GPI_PLC_R $GPI_PLC_P)

MiniTest_F="TTAGTCGTAAGTTGCAAGAAG|$(./complement.sh TTAGTCGTAAGTTGCAAGAAG)"
MiniTest_R="TTTCCACATCAAACTACCTTC|$(./complement.sh TTTCCACATCAAACTACCTTC)"
MiniTest_P="TATTATGGATGGTTATAAAAGGATA|$(./complement.sh TATTATGGATGGTTATAAAAGGATA)"

all_probes=($MiniTest_F $MiniTest_R $MiniTest_P)

source <(aenv --shell bash --data samples.csv --no_sniffer)


sample="${name##*/}"

count () {
  echo "awk '"
  for p in ${!all_probes[@]}; do echo "/${all_probes[$p]}/{c[$p]++}" ; done
  echo "END { printf \"%s\", \"$sample\""
  for p in ${!all_probes[@]} ; do echo "; printf \" %d\", c[$p]" ; done
  echo "; printf \"\n\" }"
  echo "'"
}

gzip -cd rawdata/$sample/*.gz | eval $(count) > $PBS_JOBNAME.tmp 

exit_code=$?

alog --state end  --exit "$exit_code" 

#echo $sample $(cat $sample.tmp.[1-2] | tr '\n' ' ') > $PBS_JOBNAME-$PBS_ARRAYID.tmp
#rm $sample.tmp.[1-2]