Automation using SGE

launch_downsampling.sh

+#! /bin/sh
+
+### ASDP PIPELINE ###
+## Licence: AGPLV3
+## Author: Anne-Sophie Denommé-Pichon
+## Description: script to launch the wrapper to downsample data
+
+#ROOTOUTPUTDIR: downsampled BAM output directory
+ROOTOUTPUTDIR="/work/gad/shared/analyse/STR/Data/downsampling"
+
+#EXPERIMENT_COUNT: number of randomized tests the user want for each sample and for each downsampling rate
+EXPERIMENT_COUNT=20
+
+COMPUTE_QUEUE="batch"
+
+for inputfile in \
+    "/work/gad/shared/analyse/STR/Data/dijen017/dijen017/dijen017.bam" \
+    "/work/gad/shared/analyse/STR/Data/dijen402/dijen402.bam"
+do
+    sample="$(basename -s .bam "$inputfile")"
+    for rate in 5 10 15 20 25 30 35 40 # Downsampling rate: 0 - 100
+    do
+	for experiment_number in $(seq $EXPERIMENT_COUNT)
+	do
+	    outputdir="${ROOTOUTPUTDIR}/${sample}_downsampling_${rate}_${experiment_number}"
+	    mkdir -p "$outputdir"
+	    qsub -pe smp 1 -o "$outputdir" -e "$outputdir" -q "$COMPUTE_QUEUE" \
+		-v INPUTFILE="$inputfile",OUTPUTDIR="$outputdir",SAMPLE="$sample",DOWNSAMPLING_RATE="$rate",EXPERIMENT_NUMBER="$experiment_number",LOGFILE="$outputdir/${sample}_downsampling_wrapper_${rate}_${experiment_number}.$(date +"%F_%H-%M-%S").log" "$(dirname "$0")/wrapper_downsampling.sh"
+	done
+    done
+done

wrapper_downsampling.sh

 #! /bin/sh
 
-DIJEN="dijen017"
-INPUT="/work/gad/shared/analyse/STR/Data/${DIJEN}/${DIJEN}/${DIJEN}.bam"
-DSDIR="/work/gad/shared/analyse/STR/Data/downsampling"
-DSR=10
-
-#DSR : downsampling rate (0 - 100)
-#DSRDIR : downsampled BAM directory
+### ASDP PIPELINE ###
+## Licence: AGPLv3
+## Author: Anne-Sophie Denommé-Pichon
+## Description: a wrapper for BAM downsampling
+## Usage: qsub -pe smp 1 -v INPUTFILE=<path to the FASTQ file>,OUTPUTDIR=<output directory>,SAMPLE=<dijen or dijex>,DOWNSAMPLING_RATE=<percentage of reads to keep in the BAM file>,EXPERIMENT_NUMBER=<experiment number>,[LOGFILE=<path to the log file>] wrapper_downsampling.sh
 
 set -x
 
-mkdir -p "${DSDIR}/${DIJEN}_downsampling_${DSR}_${i}"
+# Log file path option
+if [ -z "$LOGFILE" ]
+then
+    echo "Logfile is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+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 "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
+
+# Check if output directory is specified
+if [ -z "$OUTPUTDIR" ]
+then
+    echo "Output prefix is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
+
+# Check if the sample is specified
+if [ -z "$SAMPLE" ]
+then
+    echo "Sample is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
+
+# Check if downsampling rate is specified
+if [ -z "$DOWNSAMPLING_RATE" ]
+then
+    echo "Downsampling rate is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
 
-/user1/gad/an1770de/Scripts/bam_downsampling.py \
-   -i "${INPUT}" \
-   -o "${DSDIR}/${DIJEN}_downsampling_${DSR}_${i}/${DIJEN}_downsampling_${DSR}_${i}.bam" \
+# Check if experiment number is specified
+if [ -z "$EXPERIMENT_NUMBER" ]
+then
+    echo "Experiment number is not specified"
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
+
+DSR="$DOWNSAMPLING_RATE"
+I="$EXPERIMENT_NUMBER"
+
+# Command
+"$(dirname "$0")/bam_downsampling.py" \
+   -i "${INPUTFILE}" \
+   -o "${OUTPUTDIR}/${SAMPLE}_downsampling_${DSR}_${I}.bam" \
    -d "${DSR}" \
-   -l "${DSDIR}/${DIJEN}_downsampling_${DSR}_${i}/${DIJEN}_downsampling_${DSR}_${i}.log"
+   -l "${OUTPUTDIR}/${SAMPLE}_downsampling_${DSR}_${I}.log"
+
+samtools index "${OUTPUTDIR}/${SAMPLE}_downsampling_${DSR}_${I}.bam"
+
+downsampling_exitcode=$?
+
+# Check exit code
+echo "downsampling exit code : $downsampling_exitcode"
+if [ $downsampling_exitcode != 0 ]
+then
+    echo "$(date +"%F_%H-%M-%S"): END"
+    exit 1
+fi
 
-samtools index "${DSDIR}/${DIJEN}_downsampling_${DSR}_${i}/${DIJEN}_downsampling_${DSR}_${i}.bam"
+echo "$(date +"%F_%H-%M-%S"): END"