Merge pull request #40 from arahuja/referenceend

Fixing reference position from offset implementation
bigdatagenomics · Jan 10, 2014 · 1977ba4 · 1977ba4
2 parents df838ea + df16e01
commit 1977ba4
Show file tree

Hide file tree

Showing 2 changed files with 64 additions and 1 deletion.
diff --git a/adam-core/src/main/scala/edu/berkeley/cs/amplab/adam/rich/RichADAMRecord.scala b/adam-core/src/main/scala/edu/berkeley/cs/amplab/adam/rich/RichADAMRecord.scala
@@ -148,7 +148,7 @@ class RichADAMRecord(record: ADAMRecord) {
                CigarOperator.EQ |
                CigarOperator.S => {
             val positions = Range(posAtCigar.toInt, posAtCigar.toInt + elem.getLength)
-            (positions.last, basePositions ++ positions.map(t => Some(t.toLong)))
+            (positions.last + 1, basePositions ++ positions.map(t => Some(t.toLong)))
           }
           case CigarOperator.H => {
             runningPos /* do nothing */

diff --git a/adam-core/src/test/scala/edu/berkeley/cs/amplab/adam/rich/RichADAMRecordSuite.scala b/adam-core/src/test/scala/edu/berkeley/cs/amplab/adam/rich/RichADAMRecordSuite.scala
@@ -41,6 +41,19 @@ class RichADAMRecordSuite extends FunSuite {
     assert(recordWithHardClipping.unclippedEnd == Some(20L))
   }
 
+  test( "Reference End Position")
+  {
+    val unmappedRead = ADAMRecord.newBuilder().setReadMapped(false).setStart(0).setCigar("10M").build()
+    val recordWithoutClipping = ADAMRecord.newBuilder().setReadMapped(true).setCigar("10M").setStart(10).build()
+    val recordWithClipping = ADAMRecord.newBuilder().setReadMapped(true).setCigar("8M2S").setStart(10).build()
+    val recordWithHardClipping = ADAMRecord.newBuilder().setReadMapped(true).setCigar("6M2S2H").setStart(10).build()
+    assert(unmappedRead.end == None)
+    assert(recordWithoutClipping.end == Some(20L))
+    assert(recordWithClipping.end == Some(18L))
+    assert(recordWithHardClipping.end == Some(16L))
+
+  }
+
   test("Illumina Optics") {
     val nonIlluminaRecord = ADAMRecord.newBuilder().setReadName("THISISNOTILLUMINA").build()
     assert(nonIlluminaRecord.illuminaOptics == None)
@@ -54,4 +67,54 @@ class RichADAMRecordSuite extends FunSuite {
     }
   }
 
+  test("Cigar Clipping Sequence") {
+    val softClippedRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(100).setCigar("10S90M").build()
+    assert( softClippedRead.referencePositions(0) == Some(90L))
+
+
+  }
+
+  test("Reference Positions") {
+
+    val hardClippedRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("90M10H").build()
+    assert( hardClippedRead.referencePositions.length == 90)
+    assert( hardClippedRead.referencePositions(0) == Some(1000L))
+
+    val softClippedRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("10S90M").build()
+    assert( softClippedRead.referencePositions.length == 100)
+    assert( softClippedRead.referencePositions(0) == Some(990L))
+    assert( softClippedRead.referencePositions(10) == Some(1000L))
+
+    val doubleMatchNonsenseRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("10M10M").build()
+    Range(0,20).foreach(  i => assert( doubleMatchNonsenseRead.referencePositions(i) == Some(1000 + i) ))
+
+    val deletionRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("5M5D10M").build()
+    assert( deletionRead.referencePositions.length == 15)
+    assert( deletionRead.referencePositions(0) == Some(1000L))
+    assert( deletionRead.referencePositions(5) == Some(1010L))
+
+    val insertionRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("10M2I10M").build()
+    assert( insertionRead.referencePositions.length == 22)
+    assert( insertionRead.referencePositions(0) == Some(1000L))
+    assert( insertionRead.referencePositions(10) == None)
+    assert( insertionRead.referencePositions(12) == Some(1010L))
+
+    val indelRead = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("10M3D10M2I").build()
+    assert( indelRead.referencePositions.length == 22)
+    assert( indelRead.referencePositions(0) == Some(1000L))
+    assert( indelRead.referencePositions(10) == Some(1013L))
+    assert( indelRead.referencePositions(20)  == None )
+
+    val hg00096read = ADAMRecord.newBuilder().setReadMapped(true).setStart(1000).setCigar("1S28M1D32M1I15M1D23M").build()
+    assert( hg00096read.referencePositions.length == 100)
+    assert( hg00096read.referencePositions(0) == Some(999L))
+    assert( hg00096read.referencePositions(1) == Some(1000L))
+    assert( hg00096read.referencePositions(29)  == Some(1029L) )
+    assert( hg00096read.referencePositions(61)  == None )
+    assert( hg00096read.referencePositions(62)  == Some(1061L) )
+    assert( hg00096read.referencePositions(78)  == Some(1078L) )
+    assert( hg00096read.referencePositions(99)  == Some(1099L) )
+
+  }
+
 }