From 23e7f1bed9e395d0a41de659fd10d8d398dc2b82 Mon Sep 17 00:00:00 2001
From: Khalid Shakir <kshakir@broadinstitute.org>
Date: Wed, 8 Feb 2012 02:12:16 -0500
Subject: [PATCH] When an interval list specifies overlapping intervals merge
 them before scattering.

---
 .../queue/extensions/gatk/GATKIntervals.scala  |  4 +++-
 .../gatk/GATKIntervalsUnitTest.scala           | 18 ++++++++++++++++--
 2 files changed, 19 insertions(+), 3 deletions(-)

diff --git a/public/scala/src/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervals.scala b/public/scala/src/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervals.scala
index 9e47f64a1..b0483f0bb 100755
--- a/public/scala/src/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervals.scala
+++ b/public/scala/src/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervals.scala
@@ -26,7 +26,7 @@ package org.broadinstitute.sting.queue.extensions.gatk
 
 import java.io.File
 import collection.JavaConversions._
-import org.broadinstitute.sting.utils.interval.IntervalUtils
+import org.broadinstitute.sting.utils.interval.{IntervalMergingRule, IntervalUtils}
 import org.broadinstitute.sting.gatk.datasources.reference.ReferenceDataSource
 import net.sf.samtools.SAMFileHeader
 import java.util.Collections
@@ -51,6 +51,8 @@ case class GATKIntervals(reference: File, intervals: List[String]) {
         IntervalUtils.parseIntervalArguments(parser, intervals)
     Collections.sort(parsedLocs)
     Collections.unmodifiableList(parsedLocs)
+    val mergedLocs = IntervalUtils.mergeIntervalLocations(parsedLocs, IntervalMergingRule.OVERLAPPING_ONLY)
+    Collections.unmodifiableList(mergedLocs)
   }
 
   lazy val contigs = locs.map(_.getContig).distinct.toList
diff --git a/public/scala/test/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervalsUnitTest.scala b/public/scala/test/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervalsUnitTest.scala
index 38abe24ef..5383b3716 100644
--- a/public/scala/test/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervalsUnitTest.scala
+++ b/public/scala/test/org/broadinstitute/sting/queue/extensions/gatk/GATKIntervalsUnitTest.scala
@@ -32,6 +32,7 @@ import org.broadinstitute.sting.gatk.datasources.reference.ReferenceDataSource
 import org.broadinstitute.sting.utils.fasta.CachingIndexedFastaSequenceFile
 import org.broadinstitute.sting.utils.{GenomeLocSortedSet, GenomeLocParser}
 import collection.JavaConversions._
+import org.broadinstitute.sting.utils.interval.IntervalUtils
 
 class GATKIntervalsUnitTest {
   private final lazy val hg18Reference = new File(BaseTest.hg18Reference)
@@ -57,7 +58,7 @@ class GATKIntervalsUnitTest {
 //    Assert.assertEquals(gi.getSplits(3).toList, List(1, 2, 3))
   }
 
-  @Test(timeOut = 30000)
+  @Test(timeOut = 30000L)
   def testIntervalFile() {
     var gi = new GATKIntervals(hg19Reference, List(BaseTest.hg19Intervals))
     Assert.assertEquals(gi.locs.size, 189894)
@@ -65,7 +66,7 @@ class GATKIntervalsUnitTest {
     //   for(Item item: javaConvertedScalaList)
     // This for loop is actually an O(N^2) operation as the iterator calls the
     // O(N) javaConvertedScalaList.size() for each iteration of the loop.
-    //Assert.assertEquals(gi.getSplits(gi.locs.size).size, 189894)
+    Assert.assertEquals(IntervalUtils.splitFixedIntervals(gi.locs, 189894).size(), 189894)
     Assert.assertEquals(gi.contigs.size, 24)
   }
 
@@ -84,4 +85,17 @@ class GATKIntervalsUnitTest {
     Assert.assertEquals(new GATKIntervals(hg18Reference, List("chr1", "chr2", "chr3")).contigs, List("chr1", "chr2", "chr3"))
     Assert.assertEquals(new GATKIntervals(hg18Reference, List("chr1:1-2", "chr1:4-5", "chr2:1-1", "chr3:2-2")).contigs, List("chr1", "chr2", "chr3"))
   }
+
+  @Test
+  def testSortAndMergeIntervals() {
+    testSortAndMergeIntervals(Seq("chr1:1-10", "chr1:1-10", "chr1:1-10"), Seq("chr1:1-10"))
+    testSortAndMergeIntervals(Seq("chr1:1-10", "chr1:1-11", "chr1:1-12"), Seq("chr1:1-12"))
+    testSortAndMergeIntervals(Seq("chr1:1-10", "chr1:11-20", "chr1:21-30"), Seq("chr1:1-10", "chr1:11-20", "chr1:21-30"))
+    testSortAndMergeIntervals(Seq("chr1:1-10", "chr1:10-20", "chr1:21-30"), Seq("chr1:1-20", "chr1:21-30"))
+    testSortAndMergeIntervals(Seq("chr1:1-10", "chr1:21-30", "chr1:10-20"), Seq("chr1:1-20", "chr1:21-30"))
+  }
+
+  private def testSortAndMergeIntervals(actual: Seq[String], expected: Seq[String]) {
+    Assert.assertEquals(new GATKIntervals(hg18Reference, actual.toList).locs.toSeq, expected.map(hg18GenomeLocParser.parseGenomeLoc(_)))
+  }
 }