From a8e9aa83c088456309e45b523902433e01847930 Mon Sep 17 00:00:00 2001
From: jlac <justin.lack@nih.gov>
Date: Wed, 30 Jan 2019 09:30:26 -0500
Subject: [PATCH] moved sort out of STAR and into samtools rule to deal with
 memory bug

---
 Rules/samtools_sort.rl | 8 ++++++++
 Rules/star.align.2.rl  | 4 ++--
 cluster.json           | 5 +++++
 rules.json             | 1 +
 4 files changed, 16 insertions(+), 2 deletions(-)
 create mode 100644 Rules/samtools_sort.rl

diff --git a/Rules/samtools_sort.rl b/Rules/samtools_sort.rl
new file mode 100644
index 0000000..a4471da
--- /dev/null
+++ b/Rules/samtools_sort.rl
@@ -0,0 +1,8 @@
+rule samtools_sort:
+     input:  "{x}.p2Aligned.out.bam",
+     output: temp("{x}.p2Aligned.sortedByCoord.out.bam")
+     params: novosort=config['bin'][pfamily]['NOVOSORT'],rname="pl:sort"
+     threads: 8
+     shell:  "module load samtools/1.9; samtools sort -@ {threads} -o {output} {input};"
+
+
diff --git a/Rules/star.align.2.rl b/Rules/star.align.2.rl
index ac0d542..1324db8 100755
--- a/Rules/star.align.2.rl
+++ b/Rules/star.align.2.rl
@@ -1,6 +1,6 @@
 rule star_align_2:
      input:  file1="{x}.R1.trimmed.fastq.gz",file2="{x}.R2.trimmed.fastq.gz",length="QC/{x}_readlength.txt",tab=expand("{x}SJ.out.tab",x=samples)
-     output: out1=temp("{x}.p2Aligned.sortedByCoord.out.bam"),out4="{x}.p2SJ.out.tab",out5="{x}.p2Log.final.out"
+     output: out1=temp("{x}.p2Aligned.out.bam"),out4="{x}.p2SJ.out.tab",out5="{x}.p2Log.final.out"
      params: rname='pl:star2p',prefix="{x}.p2",outsamunmapped=config['bin'][pfamily]['OUTSAMUNMAPPED'],wigtype=config['bin'][pfamily]['WIGTYPE'],wigstrand=config['bin'][pfamily]['WIGSTRAND'], gtffile=config['references'][pfamily]['FUSIONGTFFILE'], nbjuncs=config['bin'][pfamily]['NBJUNCS'],starref=config['references'][pfamily]['STARREF']
      threads: 32
      run:
@@ -8,5 +8,5 @@ rule star_align_2:
         rl=int(open(input.length).readlines()[0].strip())-1
         dbrl=sorted(list(map(lambda x:int(re.findall("genes-(\d+)",x)[0]),glob.glob(params.starref+'*/',recursive=False))))
         bestdbrl=next(x[1] for x in enumerate(dbrl) if x[1] >= rl)
-        cmd="module load STAR/2.5.2b; STAR --genomeDir {params.starref}"+str(bestdbrl)+" --readFilesIn {input.file1} {input.file2} --readFilesCommand zcat --runThreadN {threads} --outFileNamePrefix {params.prefix} --outSAMunmapped {params.outsamunmapped} --sjdbFileChrStartEnd {input.tab} --sjdbGTFfile {params.gtffile} --limitSjdbInsertNsj 10000000 --outSAMtype BAM SortedByCoordinate --limitBAMsortRAM 39627002904"
+        cmd="module load STAR/2.5.2b; STAR --genomeDir {params.starref}"+str(bestdbrl)+" --readFilesIn {input.file1} {input.file2} --readFilesCommand zcat --runThreadN {threads} --outFileNamePrefix {params.prefix} --outSAMunmapped {params.outsamunmapped} --sjdbFileChrStartEnd {input.tab} --sjdbGTFfile {params.gtffile} --limitSjdbInsertNsj 10000000 --outSAMtype BAM Unsorted"
         shell(cmd)
\ No newline at end of file
diff --git a/cluster.json b/cluster.json
index ebbfa13..8947381 100755
--- a/cluster.json
+++ b/cluster.json
@@ -24,6 +24,11 @@
 		"gres": "lscratch:256",
 		"threads": "2"
 	},
+	"samtools_sort": {
+		"mem": "48g",
+		"gres": "lscratch:256",
+		"threads": "8"
+	},
 	"picard_dedup": {
 		"mem": "96g",
 		"gres": "lscratch:256",
diff --git a/rules.json b/rules.json
index 11357ad..c40dcba 100755
--- a/rules.json
+++ b/rules.json
@@ -7,6 +7,7 @@
         "freec_wgs_tumoronly": ["wgs-somatic-tumoronly"],
         "sequenza": ["wgs-somatic","exomeseq-somatic"],
         "avia": ["wgslow","exomeseq-germline","exomeseq-germline-recal","exomeseq-germline-partial","rnaseqvargerm"],
+        "samtools_sort": ["rnaseqvargerm"],
         "index_bams": ["wgslow","exomeseq-germline","wgs-somatic-tumoronly","exomeseq-somatic-tumoronly","wgs-somatic","exomeseq-somatic"],
         "rnaseqforfusions": ["rnaseqfusion"],
         "rnafusioncleanup": ["none"],