In-process versions of sample extraction and interval-list conversion for VCF files. Required an in-process-function branch of the queue library.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@4827 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

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

@@ -0,0 +1,31 @@
+package org.broadinstitute.sting.queue.library.ipf.vcf
+
+import collection.JavaConversions._
+import org.broadinstitute.sting.queue.function.InProcessFunction
+import org.broadinstitute.sting.commandline._
+import org.broadinstitute.sting.utils.text.XReadLines
+import java.io.{PrintStream, PrintWriter, File}
+
+class VCFExtractIntervals(inVCF: File, outList: File, useFilterSites: Boolean) extends InProcessFunction {
+  def this(in : File, out: File) = this(in,out,true)
+  def this(in : File) = this(in,new File(in.getAbsolutePath.replace(".vcf",".intervals.list")),true)
+
+  @Input(doc="The VCF to convert to an interval list") var vcfIn : File = inVCF
+  @Output(doc="The intervals file to write to") var listOut : File = outList
+  @Argument(doc="Keep filtered sites?") var keepFilters : Boolean = useFilterSites
+
+  var out : PrintWriter = _
+
+  def run = {
+    out = new PrintWriter(new PrintStream(listOut))
+    asScalaIterator(new XReadLines(vcfIn)).foreach(vcf2int)
+  }
+
+  def vcf2int( vcfLine: String ) : Unit = {
+    var spline = vcfLine.split("\t")
+    if ( ! vcfLine.startsWith("#") && (spline(6).equals("PASS") || keepFilters) ) {
+      out.print("%s:%s%n".format(spline(0),spline(1)))
+    }
+  }
+
+}

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

@@ -0,0 +1,36 @@
+package org.broadinstitute.sting.queue.library.ipf.vcf
+
+import org.broadinstitute.sting.queue.function.InProcessFunction
+import org.broadinstitute.sting.utils.text.XReadLines
+import collection.JavaConversions._
+import org.broadinstitute.sting.commandline._
+import java.io.{PrintWriter, PrintStream, File}
+
+class VCFExtractSamples(inVCF: File, outVCF: File, samples: List[String]) extends InProcessFunction {
+  def this(in: File, out: File, samples: File) = this(in,out, (new XReadLines(samples)).readLines.toList)
+
+  @Input(doc="VCF from which to extract samples") var inputVCF : File = inVCF
+  @Output(doc="VCF to which to write the sample-subset vcf") var outputVCF : File = outVCF
+  @Argument(doc="The samples to extract from the VCF") var extractSamples : List[String] = samples
+
+  var out : PrintWriter = _
+  var columns : List[Int] = 0 to 8 toList
+
+  def run = {
+    out = new PrintWriter(new PrintStream(outputVCF))
+    asScalaIterator(new XReadLines(inputVCF)).foreach(subset)
+  }
+
+  def subset( line : String ) {
+    if ( line.startsWith("##") ) {
+      out.print("%s%n".format(line))
+    } else {
+      val spline = line.split("\t")
+      if ( spline(0).equals("#CHROM") ) {
+        columns ++= spline.zipWithIndex.filter( p => samples.contains(p._1) ).map( p => p._2 )
+      }
+
+      out.print("%s%n".format(columns.map(p => spline(p)).reduceLeft(_ + "\t" + _)))
+    }
+  }
+}