Commit for Khalid -- should be a scala version of vcf2table but for some reason the run method isn't getting called.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4841 348d0f76-0448-11de-a6fe-93d51630548a

scala/qscript/chartl/private_mutations.q

@@ -2,7 +2,7 @@ import collection.JavaConversions._
 import java.io.FileNotFoundException
 import org.broadinstitute.sting.datasources.pipeline._
 import org.broadinstitute.sting.queue.extensions.gatk._
-import org.broadinstitute.sting.queue.library.clf.vcf._
+import org.broadinstitute.sting.queue.library.ipf.vcf._
 import org.broadinstitute.sting.queue.pipeline._
 import org.broadinstitute.sting.queue.QScript
 import org.broadinstitute.sting.utils.yaml.YamlUtils
@@ -63,7 +63,7 @@ class private_mutations extends QScript {
     eval_all.out = swapExt(finalMergedVCF,".vcf",".perm.csv")
     eval_all.reportType = Some(org.broadinstitute.sting.utils.report.VE2ReportFactory.VE2TemplateType.CSV)
 
-    //add(eval_all)
+    add(eval_all)
 
     var eval_afr : VariantEval = vcLib.addTrait(new VariantEval)
     eval_afr.rodBind :+= new RodBind("evalAFR","VCF",extract_afr.outputVCF)
@@ -73,7 +73,7 @@ class private_mutations extends QScript {
     eval_afr.reportType = Some(org.broadinstitute.sting.utils.report.VE2ReportFactory.VE2TemplateType.CSV)
     eval_afr.noStandard = true
 
-    //add(eval_afr)
+    add(eval_afr)
 
     var eval_eur : VariantEval = vcLib.addTrait(new VariantEval)
     eval_eur.rodBind :+= new RodBind("compAFR","VCF",extract_afr.outputVCF)
@@ -83,7 +83,7 @@ class private_mutations extends QScript {
     eval_eur.reportType = Some(org.broadinstitute.sting.utils.report.VE2ReportFactory.VE2TemplateType.CSV)
     eval_eur.noStandard = true
 
-    //add(eval_eur)
+    add(eval_eur)
   }
   
-}
+}

scala/qscript/lib/Vcf2Table.q

@@ -0,0 +1,22 @@
+import org.broadinstitute.sting.commandline.Hidden
+import org.broadinstitute.sting.queue.QScript
+
+import org.broadinstitute.sting.queue.library.ipf.vcf.VCFInfoToTable
+import collection.JavaConversions._
+
+class Vcf2Table extends QScript {
+  @Argument(shortName="vcf",doc="VCF file",required=true) var vcf : File = _
+  @Argument(shortName="f",doc="Info fields to extract",required=false) var fields : java.util.List[String] = new java.util.ArrayList[String]
+  @Argument(shortName="o",doc="Output file",required=true) var output : File = _
+  @Argument(shortName="useFilters",doc="Use filtered sites?",required=false) var useFilters : Boolean = false
+  @Hidden @Argument(shortName="pass",doc="set the hack filter string to this value",required=false) var filterString : String = "PASS"
+  @Hidden @Argument(shortName="notfound",doc="set the hack no entry string to this value",required=false) var keyNotFound : String = "NA"
+
+  def script = {
+    var vcf2table : VCFInfoToTable = new VCFInfoToTable(vcf,output,fields,useFilters)
+    vcf2table.PF_KEY = filterString
+    vcf2table.NO_KEY = keyNotFound
+    add(vcf2table)
+    this.functions.foreach(u => logger.debug("added: %s%n".format(u.toString)))
+  }
+}

scala/src/org/broadinstitute/sting/queue/library/ipf/vcf/VCFInfoToTable.scala

@@ -0,0 +1,56 @@
+package org.broadinstitute.sting.queue.library.ipf.vcf
+
+import org.broadinstitute.sting.commandline._
+import org.broadinstitute.sting.utils.text.XReadLines
+import collection.JavaConversions._
+import java.io.{PrintStream, PrintWriter, File}
+import collection.immutable.HashSet
+import collection.mutable.HashMap
+import org.broadinstitute.sting.queue.function.InProcessFunction
+
+class VCFInfoToTable(vcf: File, table: File, annots: List[String], keepFilter: Boolean) extends InProcessFunction {
+  def this(in: File, out: File, anns: List[String]) = this(in,out,anns,true)
+  def this(in: File, out: File) = this(in,out,Nil,true)
+  def this(in: File) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")), Nil, true)
+  def this(in: File, anns: List[String]) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns,true)
+  def this(in: File, anns: java.util.List[String]) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns.toList,true)
+  def this(in: File, anns: List[String], keep: Boolean) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns,keep)
+  def this(in: File, anns: java.util.List[String], keep: Boolean) = this(in, new File(in.getAbsolutePath.replace(".vcf",".info.table")),anns.toList,keep)
+  def this(in: File, out: File, anns: java.util.List[String], keep: Boolean) = this(in,out,anns.toList,keep)
+
+  @Input(doc="VCF file from which to extract annotion") var inVCF: File = vcf
+  @Output(doc="Table file to which to write") var outTable: File = table
+  @Argument(doc="Annotations to extract from info field") var annotations: List[String] = annots
+  @Argument(doc="Keep filtered records?") var keepFilteredRecs: Boolean = keepFilter
+
+  // set as vars so pipelnes can hack these values
+  var NO_KEY : String = "NA"
+  var PF_KEY : String = "PASS"
+
+  var out : PrintWriter = _
+  var annotation_set : HashSet[String] = new HashSet[String]
+
+  //todo -- Khalid: Why is run not being called?
+  def run = {
+    logger.debug("RUN IS CALLED")
+    annotation_set ++= annotations
+    out = new PrintWriter(new PrintStream(outTable))
+    asScalaIterator(new XReadLines(inVCF)).foreach(lineToTable)
+  }
+
+  def lineToTable(line : String) = {
+    if ( ! line.startsWith("#") ) {
+      val spline = line.split("\t")
+      if ( spline(6).equals(PF_KEY) || keepFilteredRecs ) {
+        val iMap = spline(7).split(";").map(_.split("=")).filter(p => annotation_set.contains(p.apply(0))).foldLeft(new HashMap[String,String])( (a,b) => a += new Tuple2(b.apply(0),b.apply(1)) )
+        out.print("%s%n".format(annotations.map( u => {
+          if ( iMap.contains(u) ) {
+            iMap.get(u)
+          } else {
+            NO_KEY
+          }
+        }).reduceLeft((a,b) => a + "\t" + b)))
+      }
+    }
+  }
+}