From fae0fb58ec103c13c13026f1850a9ff0a24fdb17 Mon Sep 17 00:00:00 2001
From: Daniel Cameron <d-cameron@github.com>
Date: Tue, 20 Oct 2020 12:33:59 +1100
Subject: [PATCH] #410 added locking to setupreference step switched to always
 creating {REFERENCE}.dict during sequence dictionary check

---
 scripts/gridss.sh                             | 39 ++++++++++++++-----
 .../cmdline/ReferenceCommandLineProgram.java  |  4 +-
 2 files changed, 32 insertions(+), 11 deletions(-)

diff --git a/scripts/gridss.sh b/scripts/gridss.sh
index 260619b45..6a236240a 100644
--- a/scripts/gridss.sh
+++ b/scripts/gridss.sh
@@ -518,23 +518,44 @@ if [[ "$useproperpair" == "true" ]] ; then
 fi
 
 if [[ $do_setupreference == true ]] ; then
+	lock_file=$reference.tmp.gridsslock
 	if [[ ! -f ${reference}.fai ]] && [[ ! -f $(basename $reference .fa).fai ]] && [[ ! -f $(basename $reference .fasta).fai ]]  ; then
-		write_status "Running	samtools faidx	(once-off setup for reference genome)"
-		$timecmd samtools faidx $reference 1>&2 2>> $logfile
+		if mkdir $lock_file ; then
+			write_status "Running	samtools faidx	(once-off setup for reference genome)"
+			$timecmd samtools faidx $reference 1>&2 2>> $logfile
+			rmdir $lock_file
+		else
+			write_status "Multiple instances of setupreference cannot be run concurrently. Aborting due to existence of $lock_file"
+			exit $EX_CANTCREAT
+		fi
 	fi
 	if [[ ! -f ${reference}.bwt  ]] ; then
-		write_status "Running	bwa index	(once-off setup for reference genome)"
-		$timecmd bwa index $reference 1>&2 2>> $logfile
+		if mkdir $lock_file ; then
+			write_status "Running	bwa index	(once-off setup for reference genome)"
+			$timecmd bwa index $reference 1>&2 2>> $logfile
+			rmdir $lock_file
+		else
+			write_status "Multiple instances of setupreference cannot be run concurrently. Aborting due to existence of $lock_file"
+			exit $EX_CANTCREAT
+		fi
 	fi
 	#if [[ ! -f ${reference}.idx  ]] ; then
 	#	write_status "Running	minimap2 index	(once-off setup for reference genome)"
 	#	$timecmd minimap2 -d ${reference}.idx ${reference} 1>&2 2>> $logfile
 	#fi
-	write_status "Running	PrepareReference	(once-off setup for reference genome)"
-	$timecmd java -Xmx4g $jvm_args \
-		-cp $gridss_jar gridss.PrepareReference \
-		REFERENCE_SEQUENCE=$reference \
-		1>&2 2>> $logfile
+	if [[ ! -f ${reference}.gridsscache ]] || [[ ! -f ${reference}.img ]] || [[ ! -f ${reference}.dict ]] ; then
+		if mkdir $lock_file ; then
+			write_status "Running	PrepareReference	(once-off setup for reference genome)"
+			$timecmd java -Xmx4g $jvm_args \
+				-cp $gridss_jar gridss.PrepareReference \
+				REFERENCE_SEQUENCE=$reference \
+				1>&2 2>> $logfile
+			rmdir $lock_file
+		else
+			write_status "Multiple instances of setupreference cannot be run concurrently. Aborting due to existence of $lock_file"
+			exit $EX_CANTCREAT
+		fi
+	fi
 fi
 
 if [[ $do_preprocess == true ]] ; then
diff --git a/src/main/java/gridss/cmdline/ReferenceCommandLineProgram.java b/src/main/java/gridss/cmdline/ReferenceCommandLineProgram.java
index 3837c32ed..20d9a6619 100644
--- a/src/main/java/gridss/cmdline/ReferenceCommandLineProgram.java
+++ b/src/main/java/gridss/cmdline/ReferenceCommandLineProgram.java
@@ -48,9 +48,9 @@ public ReferenceLookup getReference() {
 	public static boolean ensureSequenceDictionary(File referenceFile) {
 		try (ReferenceSequenceFile rsf = new FastaSequenceFile(referenceFile, false)) {
 			Path path = referenceFile.toPath().toAbsolutePath();
-			if (rsf.getSequenceDictionary() == null) {
+			Path dictPath = path.resolveSibling(path.getFileName().toString() + FileExtensions.DICT);
+			if (!dictPath.toFile().exists()) {
 				log.info("Attempting to create sequence dictionary for " + referenceFile);
-				Path dictPath = path.resolveSibling(path.getFileName().toString() + FileExtensions.DICT);
 				CommandLineProgramHelper cmd = new CommandLineProgramHelper(new picard.sam.CreateSequenceDictionary());
 				cmd.addArg("OUTPUT", dictPath.toFile());
 				cmd.addArg("R", referenceFile.getPath());