From 3c1bf40a45025cd1bf56f8dbe784b69d06aeb8e3 Mon Sep 17 00:00:00 2001
From: chartl <chartl@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Wed, 23 Mar 2011 19:42:29 +0000
Subject: [PATCH] QScript for scatter-gathering regional association (not quite
 as easy as using the built-in extension, due to the multiplexer). Currently
 does not work due to something I'm missing re: scatter gather class, this
 commit is an interim one.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5500 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../oneoffs/chartl/ScatterGatherAssociation.q | 89 +++++++++++++++++++
 1 file changed, 89 insertions(+)
 create mode 100755 scala/qscript/oneoffs/chartl/ScatterGatherAssociation.q

diff --git a/scala/qscript/oneoffs/chartl/ScatterGatherAssociation.q b/scala/qscript/oneoffs/chartl/ScatterGatherAssociation.q
new file mode 100755
index 000000000..ec16611c1
--- /dev/null
+++ b/scala/qscript/oneoffs/chartl/ScatterGatherAssociation.q
@@ -0,0 +1,89 @@
+import org.broadinstitute.sting.commandline.{Argument, Output, Input}
+import org.broadinstitute.sting.queue.extensions.gatk.CommandLineGATK
+import org.broadinstitute.sting.queue.QScript
+import org.broadinstitute.sting.utils.text.XReadLines
+import collection.JavaConversions._
+
+class ScatterGatherAssociation extends QScript {
+
+  @Argument(fullName="gatkJar",shortName="gatk",doc="Path to the GATK jarfile",required=true)
+  var gatkJar : File = _
+  @Argument(fullName="metaData",shortName="SM",doc="Sample meta data",required=true)
+  var metaData : File = _
+  @Argument(fullName="bamList",shortName="I",doc="list of bam files (single .list file)",required=true)
+  var bamList : File = _
+  @Argument(fullName="outputBase",shortName="o",doc="Base for output files",required=true)
+  var outBase : String = _
+  @Argument(fullName="noBedGraph",shortName="nbg",doc="Don't use bedgraph format",required=false)
+  var dontUseBedGraph : Boolean = false
+  @Argument(fullName="reference",shortName="R",doc="Reference file, if not hg19",required=false)
+  var referenceFile : File = new File("/seq/references/Homo_sapiens_assembly19/v1/Homo_sapiens_assembly19.fasta")
+  @Argument(fullName="intervals",shortName="L",doc="Interval list, if not whole-exome 1.1",required=false)
+  var intervalsFile : File = new File("/seq/references/HybSelOligos/whole_exome_agilent_1.1_refseq_plus_3_boosters/whole_exome_agilent_1.1_refseq_plus_3_boosters.Homo_sapiens_assembly19.targets.interval_list")
+  @Argument(fullName="memoryLimit",shortName="M",doc="Memory limit for SG jobs",required=false)
+  var memLimit : Int = 4
+  @Argument(fullName="scatterJobs",shortName="SJ",doc="Number of scatter jobs",required=false)
+  var scatterJobs : Int = 75
+
+  val ASSOCIATION_TESTS = List("BaseQualityScore","InsertSizeDistribution","MappingQuality0",
+    "MateMappingQuality","MateOtherContig","MateSameStrand","MateUnmapped","MismatchRate",
+    "ProperPairs","ReadClipping","ReadIndels","ReadMappingQuality","ReferenceMismatches",
+    "SampleDepth")
+
+  class RegionalAssociationSG(base : String, ext : String) extends CommandLineGATK with ScatterGatherableFunction{
+    this.analysis_type = "RegionalAssociation"
+
+    // the rest are output files implicitly constructed by the multiplexer
+
+    @Output(doc="bqs")
+    var bqs : File = new File(String.format("%s.%s.%s", base, "BaseQualityScore", ext))
+    @Output(doc="isd")
+    var isd : File = new File(String.format("%s.%s.%s",base,"InsertSizeDistribution",ext))
+    @Output(doc="mq0")
+    var mq0 : File = new File(String.format("%s.%s.%s",base,"MappingQuality0",ext))
+    @Output(doc="mmq")
+    var mmq : File = new File(String.format("%s.%s.%s",base,"MateMappingQuality",ext))
+    @Output(doc="moc")
+    var moc : File = new File(String.format("%s.%s.%s",base,"MateOtherContig",ext))
+    @Output(doc="mss")
+    var mss : File = new File(String.format("%s.%s.%s",base,"MateSameStrand",ext))
+    @Output(doc="mu")
+    var mu : File = new File(String.format("%s.%s.%s",base,"MateUnmapped",ext))
+    @Output(doc="mmr")
+    var mmr : File = new File(String.format("%s.%s.%s",base,"MismatchRate",ext))
+    @Output(doc="pp")
+    var pp : File = new File(String.format("%s.%s.%s",base,"ProperPairs",ext))
+    @Output(doc="rc")
+    var rc : File = new File(String.format("%s.%s.%s",base,"ReadClipping",ext))
+    @Output(doc="ri")
+    var ri : File = new File(String.format("%s.%s.%s",base,"ReadIndels",ext))
+    @Output(doc="rmq")
+    var rmq : File = new File(String.format("%s.%s.%s",base,"ReadMappingQuality",ext))
+    @Output(doc="rm")
+    var rm : File = new File(String.format("%s.%s.%s",base,"ReferenceMismatches",ext))
+    @Output(doc="sd")
+    var sd : File = new File(String.format("%s.%s.%s",base,"SampleDepth",ext))
+
+    override def commandLine = super.commandLine + " -o %s".format(base)
+  }
+
+  def script = {
+
+    var ext : String = ""
+    if ( dontUseBedGraph ) {
+      ext = "tdf"
+    } else {
+      ext = "bedgraph"
+    }
+
+    var association = new RegionalAssociationSG(outBase,ext)
+    association.intervals :+= intervalsFile
+    association.reference_sequence = referenceFile
+    association.jarFile = gatkJar
+    association.input_file ++= asScalaIterable((new XReadLines(bamList)).readLines).map(u => new File(u)).toList
+    association.scatterCount = scatterJobs
+    association.memoryLimit = Some(memLimit)
+
+    add(association)
+  }
+}
\ No newline at end of file