Create pipeline to launch STR detection tools on GAD genomes cohort

launch_ehdn.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## launch_wrapper_ehdn.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : script to launch the wrapper for qsubing ExpansionHunter denovo script for STR detection
+## Usage : 
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191102
+## last revision date : 20191126
+## Known bugs : None
+
+INPUTFILE=/work/gad/shared/analyse/STR/Data/dijen017/dijen017/dijen017.bam
+DATE="$(date +"%F_%H-%M-%S")"
+OUTPUTDIR="/work/gad/shared/analyse/STR/ExpansionHunterDeNovo/$DATE"
+OUTPUTPREFIX="$OUTPUTDIR/$(basename "$INPUTFILE")_$DATE"
+LOGFILE="$OUTPUTDIR/$DATE.log"
+
+# Launch the script on local host with --local option and on SGE with qsub without the --local option
+if [ $# -eq 1 ] && [ "x$1" = x--local ]
+then
+    mkdir -p "$OUTPUTDIR"
+    INPUTFILE="$INPUTFILE" OUTPUTPREFIX="$OUTPUTPREFIX" LOGFILE="$LOGFILE" "$(dirname "$0")/wrapper_ehdn.sh"
+else 
+    mkdir -p "$OUTPUTDIR"
+    qsub -pe smp 4 -q batch -v INPUTFILE="$INPUTFILE",OUTPUTPREFIX="$OUTPUTPREFIX",LOGFILE="$LOGFILE" wrapper_ehdn.sh
+fi

launch_pipeline.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## launch_pipeline.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : script to launch the pipeline for STR detection
+## Usage : 
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191208
+## last revision date : 20191208
+## Known bugs : None
+
+/work/gad/shared/bin/parallel/parallel-20150522-1.el7.cern/bin/parallel \
+    --jobs 10 \
+    --line-buffer \
+    "$(dirname "$0")/pipeline.sh" \
+    < genomes.list

launch_transfer.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## launch_wrapper_transfer.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : script to launch the wrapper for qsubing ExpansionHunter denovo script for STR detection
+## Usage : 
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191102
+## last revision date : 20191126
+## Known bugs : None
+
+INPUTFILE="/archive/gad/shared/bam_new_genome_temp/dijen016.bam"
+DATE="$(date +"%F_%H-%M-%S")"
+OUTPUTDIR="/work/gad/shared/analyse/STR/Data"
+LOGFILE="$OUTPUTDIR/$DATE.log"
+
+# Launch the script on local host with --local option and on SGE with qsub without the --local option
+if [ $# -eq 1 ] && [ "x$1" = x--local ]
+then
+    mkdir -p "$OUTPUTDIR"
+    INPUTFILE="$INPUTFILE" OUTPUTDIR="$OUTPUTDIR" LOGFILE="$LOGFILE" "$(dirname "$0")/wrapper_transfer.sh"
+else 
+    mkdir -p "$OUTPUTDIR"
+    qsub -pe smp 1 -q transfer -v INPUTFILE="$INPUTFILE",OUTPUTDIR="$OUTPUTDIR",LOGFILE="$LOGFILE" wrapper_transfer.sh
+fi

pipeline.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## pipeline.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : script to launch the pipeline for STR detection
+## Usage : 
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191208
+## last revision date : 20191208
+## Known bugs : None
+
+SAMPLE="$1"
+
+# Check if sample is specified
+if [ -z "$SAMPLE" ]
+then
+    echo "Sample is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
+
+INPUTFILE="/archive/gad/shared/bam_new_genome_temp/$SAMPLE.bam"
+DATE="$(date +"%F_%H-%M-%S")"
+OUTPUTDIR="/work/gad/shared/analyse/STR/pipeline/$SAMPLE"
+
+# Transfer bam and bai from archive to work
+mkdir -p "$OUTPUTDIR"
+qsub -pe smp 1 -q transfer -N "transfer_$SAMPLE" -v INPUTFILE="$INPUTFILE",OUTPUTDIR="$OUTPUTDIR",LOGFILE="$OUTPUTDIR/transfer_$DATE.log" wrapper_transfer.sh
+
+INPUTFILE="$OUTPUTDIR/$SAMPLE.bam"
+
+# Launch ExpansionHunter
+mkdir -p "$OUTPUTDIR/eh"
+qsub -pe smp 4 -q batch  -N "eh_$SAMPLE" -hold_jid "transfer_$SAMPLE" -v INPUTFILE="$INPUTFILE",OUTPUTPREFIX="$OUTPUTDIR/eh/$SAMPLE",LOGFILE="$OUTPUTDIR/eh/$DATE.log" wrapper_expansionhunter.sh
+
+# Launch Tredparse
+mkdir -p "$OUTPUTDIR/tredparse"
+qsub -pe smp 4 -q batch -N "tredparse_$SAMPLE" -hold_jid "transfer_$SAMPLE" -v INPUTFILE="$INPUTFILE",OUTPUTDIR="$OUTPUTDIR/tredparse",LOGFILE="$OUTPUTDIR/tredparse/$DATE.log" wrapper_tredparse.sh
+
+# Launch GangSTR
+mkdir -p "$OUTPUTDIR/gangstr"
+qsub -pe smp 4 -q batch -N "gangstr_$SAMPLE" -hold_jid "transfer_$SAMPLE" -v INPUTFILE="$INPUTFILE",OUTPUTPREFIX="$OUTPUTDIR/gangstr/$SAMPLE",LOGFILE="$OUTPUTDIR/gangstr/$DATE.log" wrapper_gangstr.sh
+
+# Launch ehdn profile
+mkdir -p "$OUTPUTDIR/ehdn"
+qsub -pe smp 4 -q batch -N "ehdn_$SAMPLE" -hold_jid "transfer_$SAMPLE" -v INPUTFILE="$INPUTFILE",OUTPUTPREFIX="$OUTPUTDIR/ehdn/$SAMPLE",LOGFILE="$OUTPUTDIR/ehdn/$DATE.log" wrapper_ehdn.sh
+
+# Delete transfered bam and bai
+qsub -pe smp 1 -q batch -hold_jid "eh_$SAMPLE,tredparse_$SAMPLE,gangstr_$SAMPLE,ehdn_$SAMPLE" -sync y -v SAMPLE="$SAMPLE",LOGFILE="$OUTPUTDIR/delete_$DATE.log" wrapper_delete.sh

wrapper_delete.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## wrapper_delete.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : a wrapper for qsubing bam deletion for STR pipeline
+## Usage : qsub -pe smp 1 -v SAMPLE=<sample>,[LOGFILE=<path to the log file>] wrapper_delete.sh
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191208
+## last revision date : 20191208
+## Known bugs : None
+
+
+# Log file path option
+if [ -z "$LOGFILE" ]
+then
+    LOGFILE=delete.$(date +"%F_%H-%M-%S").log
+fi
+
+# Logging
+exec 1>> "$LOGFILE" 2>&1
+echo "$(date +"%F_%H-%M-%S"): START"
+
+# Check if sample is specified
+if [ -z "$SAMPLE" ]
+then
+    echo "Sample is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch delete.failed
+    exit 1
+fi
+
+# Delete and check exit code
+echo "command : rm \
+    $SAMPLE"
+rm \
+    "/work/gad/shared/analyse/STR/pipeline/$SAMPLE/$SAMPLE.bam" \
+    "/work/gad/shared/analyse/STR/pipeline/$SAMPLE/$SAMPLE.bai"
+
+delete_exitcode=$?
+
+echo "delete exit code : $delete_exitcode"
+if [ $delete_exitcode != 0 ]
+then
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch delete.failed
+    exit 1
+fi
+
+echo "$(date +"%F_%H-%M-%S"): END"

wrapper_ehdn.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## wrapper_ehdn.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : a wrapper for qsubing ExpansionHunter denovo script for STR detection
+## Usage : qsub -pe smp 1 -v INPUTFILE=<path to the bam file>,OUTPUTPREFIX=<output prefix>,[LOGFILE=<path to the log file>] wrapper_ehdn.sh
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191208
+## last revision date : 20191208
+## Known bugs : None
+
+
+# Log file path option
+if [ -z "$LOGFILE" ]
+then
+    LOGFILE=ehdn.$(date +"%F_%H-%M-%S").log
+fi
+
+# Logging
+exec 1>> "$LOGFILE" 2>&1
+echo "$(date +"%F_%H-%M-%S"): START"
+
+
+# Check if input file exists
+if [ ! -f "$INPUTFILE" ]
+then
+    echo "Input file '$INPUTFILE' does not exist"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch ehdn.failed
+    exit 1
+fi
+
+# Check if output prefix is specified
+if [ -z "$OUTPUTPREFIX" ]
+then
+    echo "Output prefix is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch ehdn.failed
+    exit 1
+fi
+
+# Create .bam and .bai symbolic links
+TMPDIR="$(mktemp -d)"
+ln -s "$INPUTFILE" "$TMPDIR/$(basename "$INPUTFILE")"
+ln -s "$(echo "$INPUTFILE" | sed 's/\.bam$/.bai/')" "$TMPDIR/$(basename "$INPUTFILE").bai"
+
+# Launch script command and check exit code
+echo "command : /work/gad/shared/bin/expansionhunterdenovo/ExpansionHunterDenovo-v0.8.0-linux_x86_64/bin/ExpansionHunterDenovo-v0.8.0 profile \
+    --reads "$TMPDIR/$(basename "$INPUTFILE")" \
+    --reference /work/gad/shared/pipeline/hg19/index/hg19_essential.fa \
+    --output-prefix $OUTPUTPREFIX \
+    --min-anchor-mapq 50 \
+    --max-irr-mapq 40"
+/work/gad/shared/bin/expansionhunterdenovo/ExpansionHunterDenovo-v0.8.0-linux_x86_64/bin/ExpansionHunterDenovo-v0.8.0 profile \
+    --reads "$TMPDIR/$(basename "$INPUTFILE")" \
+    --reference /work/gad/shared/pipeline/hg19/index/hg19_essential.fa \
+    --output-prefix "$OUTPUTPREFIX" \
+    --min-anchor-mapq 50 \
+    --max-irr-mapq 40
+
+ehdn_exitcode=$?
+
+# Remove .bam and .bai symbolic links
+rm "$TMPDIR/$(basename "$INPUTFILE")"
+rm "$TMPDIR/$(basename "$INPUTFILE").bai"
+rmdir "$TMPDIR"
+
+echo "ehdn exit code : $ehdn_exitcode"
+if [ $ehdn_exitcode != 0 ]
+then
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch ehdn.failed
+    exit 1
+fi
+
+echo "$(date +"%F_%H-%M-%S"): END"

wrapper_gangstr.sh

@@ -50,18 +50,20 @@ echo "command : /work/gad/shared/bin/gangstr/GangSTR-2.4/bin/GangSTR \
     --ref /work/gad/shared/pipeline/hg19/index/hg19_essential.fa \
     --regions /work/gad/shared/bin/gangstr/STRregions/hg19_ver13_1.bed \
     --out "$OUTPUTPREFIX" \
-    --verbose \
-    --insertmean "$INSERTMEAN" \
-    --insertsdev "$INSERTDEV""
+    --verbose
+
+#    --insertmean "$INSERTMEAN" \
+#    --insertsdev "$INSERTDEV""
 
 /work/gad/shared/bin/gangstr/GangSTR-2.4/bin/GangSTR \
     --bam "$INPUTFILE" \
     --ref /work/gad/shared/pipeline/hg19/index/hg19_essential.fa \
     --regions /work/gad/shared/bin/gangstr/STRregions/hg19_ver13_1.bed \
     --out "$OUTPUTPREFIX" \
-    --verbose \
-    --insertmean "$INSERTMEAN" \
-    --insertsdev "$INSERTDEV"
+    --verbose
+
+#    --insertmean "$INSERTMEAN" \
+#    --insertsdev "$INSERTDEV"
 
 
 gangstr_exitcode=$?

wrapper_transfer.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## wrapper_transfer.sh
+## Version : 0.0.1
+## Licence : FIXME
+## Description : a wrapper for qsubing bam transfer for STR pipeline
+## Usage : qsub -pe smp 1 -v INPUTFILE=<path to the bam file>,OUTPUTDIR=<output directory>,[LOGFILE=<path to the log file>] wrapper_transfer.sh
+## Output : FIXME
+## Requirements : FIXME
+
+## Author : anne-sophie.denomme-pichon@u-bourgogne.fr
+## Creation Date : 20191208
+## last revision date : 20191208
+## Known bugs : None
+
+
+# Log file path option
+if [ -z "$LOGFILE" ]
+then
+    LOGFILE=transfer.$(date +"%F_%H-%M-%S").log
+fi
+
+# Logging
+exec 1>> "$LOGFILE" 2>&1
+echo "$(date +"%F_%H-%M-%S"): START"
+
+
+# Check if input file exists
+if [ ! -f "$INPUTFILE" ]
+then
+    echo "Input file '$INPUTFILE' does not exist"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch transfer.failed
+    exit 1
+fi
+
+# Check if output directory is specified
+if [ -z "$OUTPUTDIR" ]
+then
+    echo "Output directory is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch transfer.failed
+    exit 1
+fi
+
+# Transfer and check exit code
+echo "command : rsync -aAX \
+    $INPUTFILE $(echo "$INPUTFILE" | sed 's/\.bam$/.bai/') \
+    $OUTPUTDIR"
+rsync -aAX \
+    "$INPUTFILE" "$(echo "$INPUTFILE" | sed 's/\.bam$/.bai/')" \
+    "$OUTPUTDIR"
+
+transfer_exitcode=$? 
+
+echo "transfer exit code : $transfer_exitcode"
+if [ $transfer_exitcode != 0 ]
+then
+    echo "$(date +"%F_%H-%M-%S"): END"
+    touch transfer.failed
+    exit 1
+fi
+
+echo "$(date +"%F_%H-%M-%S"): END"