pipeline works

Master.sh

@@ -66,7 +66,7 @@ PHARMGKB=`grep "pharmagkbase" $PHARMCONFIG | cut -f2`
 OMIM=`grep "omim" $PHARMCONFIG | cut -f2`
 HLALIST=`grep "HLAlist" $PHARMCONFIG | cut -f2`
 HEADER=`grep "header" $PHARMCONFIG | cut -f2`
-
+REF=`grep "reference" $PHARMCONFIG | cut -f2`
 
 # python path for proper execution of python
 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/
@@ -104,11 +104,11 @@ else
     fi
 
     # get the status
-    REF=`grep "reference" $STATUSFILE | cut -f2`
     HLAscan=`grep "process_HLAscan" $STATUSFILE | cut -f2`
     varcallSNP=`grep "process_varcallSNP" $STATUSFILE | cut -f2`
     varcallCNV=`grep "process_varcallCNV" $STATUSFILE | cut -f2`
     createHTML=`grep "process_createHTML" $STATUSFILE | cut -f2`
+    mergeVCF=`grep "mergeVCF" $STATUSFILE | cut -f2`
 
     if [ "$HLAscan" == "FAIL" ]
     then
@@ -139,6 +139,14 @@ else
       echo "CreateHTML step was OK"
     fi
     
+    if [ "$mergeVCF" == "FAIL" ]
+    then
+      echo "MergeVCF step failed : jumping to merge step"
+      jumpto MergeVCF
+    else
+      echo "MerfeVCF step was OK"
+    fi
+
     echo "pipeline was correctly executed, no need to relaunch a step" 
     exit 0 
 fi
@@ -172,7 +180,7 @@ echo "### Execute HLAscan ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
-    echo "Command : qsub -pe smp 1 -N HLAscan_${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 $PIPELINEBASE/haplo_scan.sh"
+    echo "Command : qsub -pe smp 2 -N HLAscan_${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 $PIPELINEBASE/haplo_scan.sh"
     qsub -pe smp 1 -N HLAscan_${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")"
@@ -208,8 +216,8 @@ echo "### Generate HTML report ###"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
-    echo "Command : qsub -pe smp 1 -N createHTML_${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/$currentSample/logs/process_varcall.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/wrapper_create_html.sh"
-    qsub -pe smp 1 -N createHTML_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v VCFFILE_SNP=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,VCFFILE_CNV=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.CNV.vcf,REPORTFILE=$ANALYSISDIR/$currentSample/$currentSample.report,BAMFILE=$ANALYSISDIR/$currentSample/$currentSample.bam,TSVFILE=$TSVFILE,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.html,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_varcall.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/wrapper_create_html.sh
+    echo "Command : qsub -pe smp 1 -hold_jid varcallCNV_${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
 done
 echo "End : $(date +"%F_%H-%M-%S")"
 
@@ -219,7 +227,7 @@ echo "### Merge VCF files"
 echo "Start : $(date +"%F_%H-%M-%S")"
 for currentSample in $samples
 do
-    echo "Command : qsub -pe smp 1 -N mergeVCF_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE1=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.CNV.vcf,INPUTFILE2=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,REF=$REF,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.merge.vcf,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_varcall.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/merge_GATK.sh"
-    qsub -pe smp 1 -N mergeVCF_${currentSample} -o $ANALYSISDIR/$currentSample/logs/ -e $ANALYSISDIR/$currentSample/logs/ -v INPUTFILE1=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.CNV.vcf,INPUTFILE2=$ANALYSISDIR/$currentSample/$currentSample.pharmacoAnnot.SNP.vcf,REF=$REF,OUTPUTFILE=$ANALYSISDIR/$currentSample/$currentSample.vcf,LOGFILE=$ANALYSISDIR/$currentSample/logs/process_varcall.$logbasename.log,CONFIGFILE=$CONFIGFILE $PIPELINEBASE/merge_GATK.sh
+    echo "Command : qsub -pe smp 1 -hold_jid varcallCNV_${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
 done
 echo "End : $(date +"%F_%H-%M-%S")"

merge_GATK.sh

@@ -81,10 +81,10 @@ fi
 
 JAVA=$(grep "javacmd" $CONFIGFILE | awk '{print $2}')
 PATHSCRIPT=$(grep "pipelinebase" $CONFIGFILE | awk '{print $2}')
-GATK=$(grep "GATKbase" $CONFIGFILE | awk '{print $2}')
+#GATK=$(grep "GATKbase" $CONFIGFILE | awk '{print $2}')
 
-echo "command : $JAVA -jar $GATK -T CombineVariants -R $REF --variant $INPUTFILE1 --variant $INPUTFILE2 -o merge.vcf -genotypeMergeOptions UNIQUIFY"
-$JAVA -jar $GATK -T CombineVariants -R $REF --variant $INPUTFILE2 --variant $INPUTFILE2 -o merge.vcf -genotypeMergeOptions UNIQUIFY > $OUTPUTFILE
+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
 
 genotypeMerge_exitcode=$?
 echo "GenotypeMerge exit code : $genotypeMerge_exitcode"

pharmconfig.tsv

@@ -3,3 +3,4 @@ pharmagkbase	/work/gad/sv347413/epilepsie/Pipeline/pharmgkb_v2.vcf
 pharmagktab	/work/gad/sv347413/epilepsie/Pipeline/tableau_pharmgkb.tsv
 omim	/work/gad/sv347413/epilepsie/Pipeline/tableau_OMIM.tsv
 header	/work/gad/sv347413/epilepsie/Pipeline/header_CNV.vcf
+reference	/work/gad/shared/pipeline/hg19/index/hg19_essential.fa