Initial commit of in-process helper functions for making the BCM more robust

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5144 348d0f76-0448-11de-a6fe-93d51630548a

scala/src/org/broadinstitute/sting/queue/library/ipf/SortByRef.scala

@@ -0,0 +1,37 @@
+package org.broadinstitute.sting.queue.library.ipf
+
+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}
+import collection.immutable.HashMap
+
+
+class SortByRef( input: File, reference: File, output: File ) extends InProcessFunction {
+  @Input(doc="The file to be sorted") var inFile: File = input
+  @Input(doc="The reference fasta index") var fai: File = reference
+  @Output(doc="The file to write the sorted file to") var outFile :  File = output
+  @Argument(doc="The character or expression that separates entries") var separator : String = "\t"
+  @Argument(doc="The position of the contig in the file (1-based)") var pos: Int = 1
+  @Argument(doc="Comment characters (lines will be ignored)") var comment: List[String] = List("#")
+
+  var contigMap: HashMap[String,Int] = new HashMap[String,Int];
+
+  def contigVal( line : String ) : Int = {
+    if ( comment.contains(line.charAt(0)) ) {
+      return -1;
+    }
+
+    if ( contigMap.size < 1 ) { // no contigs
+      ( new XReadLines(fai)).readLines.map( u => u.split("\t").head).zipWithIndex.foreach( u => contigMap += u )
+    }
+
+    return contigMap( line.split(separator)(pos-1) )
+  }
+
+  def run = {
+    var w : PrintWriter = new PrintWriter(new PrintStream(outFile))
+    ( new XReadLines(inFile) ).readLines.sortBy(contigVal).foreach( u => w.println(u) )    
+  }
+}

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

@@ -0,0 +1,39 @@
+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 VCFExtractSites( vcf: File, output: File) extends InProcessFunction {
+  @Input(doc="VCF file from which to extract sites") var inVCF: File = vcf
+  @Output(doc="Sites VCF file to write to") var outVCF: File = output
+  @Argument(doc="Keep non-PASS sites") var keepFilters: Boolean = false
+  @Argument(doc="Keep info field") var keepInfo : Boolean = true
+  @Argument(doc="Keep qual field") var keepQual : Boolean = true
+
+  def lineMap( line: String ) : String = {
+    if ( line.startsWith("##") ) { return line }
+    val spline = line.split("\t",9)
+
+    if ( spline(6) == "PASS" || keepFilters ) {
+      if ( ! keepInfo ) {
+        spline(7) = "."
+      }
+      if ( ! keepQual ) {
+        spline(5) = "."
+      }
+      return spline.slice(0,8).reduceLeft( _ + "\t" + _ )
+    }
+
+    return ""
+  }
+
+  def run {
+    var w: PrintWriter = new PrintWriter( new PrintStream(outVCF) )
+    ( new XReadLines(inVCF) ).readLines().map(lineMap).view.filter( u => u != "" ).foreach( u => w.println(u) )
+  }
+
+}