Merge pull request #400 from jlac/activeDev

 moved sort out of STAR and into samtools rule to deal with memory bug

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};"
+
+

Rules/star.align.2.rl

@@ -1,12 +1,12 @@
 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:
         import glob
         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)

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",

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"],