new method CRAMFileReader.createIndexIterator (#1120)

* added a new method CRAMFileReader.createIndexIterator get reads overlapping intervals in a cram file, analogous to BAMFileReader#createIndexIterator * part of #1112
samtools · Jun 15, 2018 · 9ea0a03 · 9ea0a03
1 parent 8b55de6
commit 9ea0a03
Show file tree

Hide file tree

Showing 2 changed files with 53 additions and 13 deletions.
diff --git a/src/main/java/htsjdk/samtools/CRAMFileReader.java b/src/main/java/htsjdk/samtools/CRAMFileReader.java
@@ -466,6 +466,32 @@ void enableFileSource(final SamReader reader, final boolean enabled) {
             iterator.setFileSource(enabled ? reader : null);
     }
 
+    /**
+     * Prepare to iterate through SAMRecords that match the intersection of the given intervals and chunk boundaries.
+     * @param intervals the intervals to restrict reads to
+     * @param contained if <code>true</code>, return records that are strictly
+     *                  contained in the intervals, otherwise return records that overlap
+     * @param filePointers file pointer pairs corresponding to chunk boundaries for the
+     *                     intervals
+     */
+    public CloseableIterator<SAMRecord> createIndexIterator(final QueryInterval[] intervals,
+                                                            final boolean contained,
+                                                            final long[] filePointers) {
+        return new CRAMIntervalIterator(intervals, contained, filePointers);
+    }
+
+    // convert queries -> merged BAMFileSpan -> coordinate array
+    private static long[] coordinatesFromQueryIntervals(BAMIndex index, QueryInterval[] queries) {
+        ArrayList<BAMFileSpan> spanList = new ArrayList<>(1);
+        Arrays.asList(queries).forEach(qi -> spanList.add(index.getSpanOverlapping(qi.referenceIndex, qi.start, qi.end)));
+        BAMFileSpan spanArray[] = new BAMFileSpan[spanList.size()];
+        for (int i = 0; i < spanList.size(); i++) {
+            spanArray[i] = spanList.get(i);
+        }
+
+        return BAMFileSpan.merge(spanArray).toCoordinateArray();
+    }
+
     private class CRAMIntervalIterator extends BAMQueryMultipleIntervalsIteratorFilter
             implements CloseableIterator<SAMRecord> {
 
@@ -475,9 +501,12 @@ private class CRAMIntervalIterator extends BAMQueryMultipleIntervalsIteratorFilt
         SAMRecord nextRec = null;
 
         public CRAMIntervalIterator(final QueryInterval[] queries, final boolean contained) {
+            this(queries, contained, coordinatesFromQueryIntervals(getIndex(), queries));
+        }
+
+        public CRAMIntervalIterator(final QueryInterval[] queries, final boolean contained, final long[] coordinates) {
             super(queries, contained);
 
-            long[] coordinates = coordinatesFromQueryIntervals(getIndex(), queries);
             if (coordinates != null && coordinates.length != 0) {
                 try {
                     unfilteredIterator = new CRAMIterator(
@@ -493,18 +522,6 @@ public CRAMIntervalIterator(final QueryInterval[] queries, final boolean contain
             }
         }
 
-        // convert queries -> merged BAMFileSpan -> coordinate array
-        private long[] coordinatesFromQueryIntervals(BAMIndex index, QueryInterval[] queries) {
-            ArrayList<BAMFileSpan> spanList = new ArrayList<>(1);
-            Arrays.asList(queries).forEach(qi -> spanList.add(mIndex.getSpanOverlapping(qi.referenceIndex, qi.start, qi.end)));
-            BAMFileSpan spanArray[] = new BAMFileSpan[spanList.size()];
-            for (int i = 0; i < spanList.size(); i++) {
-                spanArray[i] = spanList.get(i);
-            }
-
-            return BAMFileSpan.merge(spanArray).toCoordinateArray();
-        }
-
         @Override
         public void close() {
             if (unfilteredIterator != null) {

diff --git a/src/test/java/htsjdk/samtools/CRAMFileCRAIIndexTest.java b/src/test/java/htsjdk/samtools/CRAMFileCRAIIndexTest.java
@@ -304,6 +304,29 @@ public void testQueryInterval() throws IOException {
         reader.close();
     }
 
+    @Test
+    public void testQueryIntervalWithFilePointers() throws IOException {
+        CRAMFileReader reader = new CRAMFileReader(
+                new ByteArraySeekableStream(cramBytes),
+                new ByteArraySeekableStream(craiBytes),
+                source,
+                ValidationStringency.STRICT);
+        QueryInterval[] query = new QueryInterval[]{new QueryInterval(0, 1519, 1520), new QueryInterval(1, 470535, 470536)};
+        BAMFileSpan fileSpan = BAMFileReader.getFileSpan(query, reader.getIndex());
+        final CloseableIterator<SAMRecord> iterator = reader.createIndexIterator(query, false, fileSpan.toCoordinateArray());
+        Assert.assertTrue(iterator.hasNext());
+        SAMRecord r1 = iterator.next();
+        Assert.assertEquals(r1.getReadName(), "3968040");
+
+        Assert.assertTrue(iterator.hasNext());
+        SAMRecord r2 = iterator.next();
+        Assert.assertEquals(r2.getReadName(), "140419");
+
+        Assert.assertFalse(iterator.hasNext());
+        iterator.close();
+        reader.close();
+    }
+
     @BeforeTest
     public void prepare() throws IOException {
         Log.setGlobalLogLevel(Log.LogLevel.ERROR);