pipeline epilepsie

Dup_Data.sh

+#!/bin/bash
+
+if [ ! -e /work/gad/sv347413/epilepsie/vcf_tmp/$i*.vcf ]
+then
+for i in $(cat /work/gad/sv347413/HLAscan/liste_patients_epileptiques.txt)
+do
+	cp /work/gad/shared/vcf/individu/$i/*.raw.vcf /work/gad/sv347413/epilepsie/vcf_tmp/.
+done
+fi
+
+if [ ! -e /work/gad/sv347413/epilepsie/tsv_tmp/$i*.tsv ]
+then
+for i in $(cat /work/gad/sv347413/HLAscan/liste_patients_epileptiques.txt)
+do 
+	cp /work/gad/shared/vcf/individu/$i/*.cnv.annot.tsv /work/gad/sv347413/epilepsie/tsv_tmp/.
+done
+fi

Master.sh

@@ -64,7 +64,6 @@ REF=`grep "reference" $PHARMCONFIG | cut -f2`
 PYTHONPATH=$PIPELINEBASE/common:/work/gad/shared/bin/lib/python_2.7/lib/python2.7/site-packages:/work/gad/shared/bin/miniconda2/lib/python2.7/site-packages/
 export PYTHONPATH
 
-
 # Sample list
 samples=`find -L $ANALYSISDIR/ -maxdepth 1 -mindepth 1 -type d | xargs -l basename`
 
@@ -75,16 +74,18 @@ then
 fi
 find -L $ANALYSISDIR/ -maxdepth 1 -mindepth 1 -type d | xargs -l basename >> $ANALYSISDIR/batch.list
 
+HLAscan:
 # HLA genotyping
 echo "### Execute HLAscan ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
-    echo "Command : qsub -pe smp 2 -N HLAscan_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,OUTPUTFILE=$ANALYSISDIR/$currentSample/HLAscan,GeneList=$GL,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_HLAscan.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/haplo_scan.sh"
+    echo "Command : qsub -pe smp 1 -N HLAscan_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,OUTPUTFILE=$ANALYSISDIR/$currentSample/HLAscan,GeneList=$GL,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_HLAscan.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/haplo_scan.sh"
     qsub -pe smp 1 -N HLAscan_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v BAM=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,SORTIE=$ANALYSISDIR/$currentSample/HLAscan,GeneList=$GL,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_HLAscan.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/haplo_scan.sh
 done
 echo "End : $(date +"%F_%H-%M-%S")"
 
+varcallSNP:
 # Variant calling first step
 echo "### Variant calling SNP first step ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
@@ -95,6 +96,7 @@ do
 done
 echo "End : $(date +"%F_%H-%M-%S")"
 
+varcallCNV:
 # Variant calling second step
 echo "### Variant calling CNV second step ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
@@ -106,22 +108,24 @@ do
 done
 echo "End : $(date +"%F_%H-%M-%S")"
 
+createHTML:
 # Generate a HTML report_vcf
 echo "### Generate HTML report ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
     echo "Command : qsub -pe smp 1 -hold_jid varcallCNV_${currentSample},varcallSNP_${currentSample},HLAscan_${currentSample} -N createHTML_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v VCF_SNP=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,VCF_CNV=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,REPORTFILE=$ANALYSISDIR/$currentSample/HLAscan/Report,BAMFILE=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,TSVFILE=$TSVFILE,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.html,LOGFILE=$ANALYSISDIR/$currentSample/logs/create_html.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/wrapper_create_html.sh"
-    qsub -pe smp 1 -hold_jid varcallCNV_${currentSample},varcallSNP_${currentSample},HLAscan_${currentSample} -N createHTML_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v VCF_SNP=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,VCF_CNV=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,REPORTFILE=$ANALYSISDIR/$currentSample/HLAscan/Report,BAMFILE=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,TSVFILE=$TSVFILE,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.html,LOGFILE=$ANALYSISDIR/$currentSample/logs/create_html.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/wrapper_create_html.sh
+    qsub -pe smp 1 -hold_jid varcallCNV_${currentSample} -N createHTML_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v VCF_SNP=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,VCF_CNV=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,REPORTFILE=$ANALYSISDIR/$currentSample/HLAscan/Report,BAMFILE=$ANALYSISDIR/$currentSample/$currentSample.dedup.bam,TSVFILE=$TSVFILE,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.html,LOGFILE=$ANALYSISDIR/$currentSample/logs/create_html.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/wrapper_create_html.sh
 done
 echo "End : $(date +"%F_%H-%M-%S")"
 
+MergeVCF:
 #merge vcf fies
 echo "### Merge VCF files"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
     echo "Command : qsub -pe smp 1 -hold_jid varcallCNV_${currentSample},varcallSNP_${currentSample} -N mergeVCF_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE1=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,INPUTFILE2=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,REF=$REF,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.merge.vcf,LOGFILE=$ANALYSISDIR/$currentSample/logs/merge_file.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/merge_GATK.sh"
-    qsub -pe smp 1 -hold_jid varcallCNV_${currentSample},varcallSNP_${currentSample} -N mergeVCF_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE1=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,INPUTFILE2=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,REF=$REF,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.merge.vcf,LOGFILE=$ANALYSISDIR/$currentSample/logs/merge_file.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/merge_GATK.sh
+    qsub -pe smp 1 -N mergeVCF_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE1=$ANALYSISDIR/$currentSample/$currentSample.CNV.vcf,INPUTFILE2=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,REF=$REF,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.merge.vcf,LOGFILE=$ANALYSISDIR/$currentSample/logs/merge_file.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/merge_GATK.sh
 done
 echo "End : $(date +"%F_%H-%M-%S")"

VCF_tools.py

@@ -74,5 +74,6 @@ def create_header():
 ##contig=<ID=chr22,length=51304566>
 ##contig=<ID=chrX,length=155270560>
 ##contig=<ID=chrY,length=59373566>
+#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	dijex
 """
     return header

haplo_scan.sh

@@ -15,7 +15,7 @@
 
 #$ -N hlascan
 #$ -q batch
-#$ -pe smp 1
+#$ -pe smp 2
 #$ -V
 
 # Log file path option
@@ -66,7 +66,7 @@ fi
 
 SAMPLE=`basename ${BAM} | cut -d . -f1`
 RES=${SORTIE}
-GL=/work/gad/shared/analyse/test_HLA/HLAscan_v1.0/hla-ref-5gene/gene_list
+GL=/work/gad/shared/bin/HLAscan_v1.0/hla-ref-5gene/gene_list
 
 #Launch haplo_scan
 /work/gad/shared/bin/miniconda2/bin/python2.7 /work/gad/shared/bin/HLAscan_v1.0/hla-paper/haplo_scan_v4.0-sv.py $BAM $GL $RES

merge_GATK.sh

@@ -79,12 +79,13 @@ then
     exit 1
 fi
 
-JAVA=$(grep "javacmd" $CONFIGFILE | awk '{print $2}')
+#JAVA=$(grep "javacmd" $CONFIGFILE | awk '{print $2}')
+module load java64/1.8.0_162
 PATHSCRIPT=$(grep "pipelinebase" $CONFIGFILE | awk '{print $2}')
 #GATK=$(grep "GATKbase" $CONFIGFILE | awk '{print $2}')
 
-echo "command : $JAVA -jar /work/gad/shared/bin/GATK_3.8/GenomeAnalysisTK.jar -T CombineVariants -R $REF --variant $INPUTFILE1 --variant $INPUTFILE2 -o merge.vcf -genotypeMergeOptions UNIQUIFY"
-$JAVA -jar /work/gad/shared/bin/GATK_3.8/GenomeAnalysisTK.jar -T CombineVariants -R $REF --variant $INPUTFILE2 --variant $INPUTFILE2 -genotypeMergeOptions UNIQUIFY > $OUTPUTFILE
+echo "command : $JAVA -jar /work/gad/shared/bin/GATK_3.7/GenomeAnalysisTK.jar -T CombineVariants -R $REF --variant $INPUTFILE1 --variant $INPUTFILE2 -genotypeMergeOptions UNIQUIFY"
+java -jar /work/gad/shared/bin/GATK_3.7/GenomeAnalysisTK.jar -T CombineVariants -R $REF --variant $INPUTFILE1 --variant $INPUTFILE2 -o $OUTPUTFILE -genotypeMergeOptions UNIQUIFY
 
 genotypeMerge_exitcode=$?
 echo "GenotypeMerge exit code : $genotypeMerge_exitcode"