From bfe90af5e210349aa1a4f74f6b209031df17b332 Mon Sep 17 00:00:00 2001
From: hanna <hanna@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Tue, 14 Jul 2009 01:25:20 +0000
Subject: [PATCH] Some quick and dirty fixes to support querying unmapped BAM
 files.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@1228 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../datasources/simpleDataSources/SAMDataSource.java | 12 ++++++++++++
 .../gatk/iterators/MergingSamRecordIterator2.java    | 12 +++++++++++-
 2 files changed, 23 insertions(+), 1 deletion(-)

diff --git a/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java b/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
index 61781c0c6..f4ce5071c 100755
--- a/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
+++ b/java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java
@@ -177,6 +177,10 @@ public class SAMDataSource implements SimpleDataSource {
     private StingSAMIterator seekRead( ReadShard shard ) throws SimpleDataSourceLoadException {
         StingSAMIterator iter = null;
 
+        // If there are no entries in the sequence dictionary, there can't possibly be any unmapped reads.  Force state to 'unmapped'.
+        if( isSequenceDictionaryEmpty() )
+            intoUnmappedReads = true;
+
         if (!intoUnmappedReads) {
             if (lastReadPos == null) {
                 lastReadPos = GenomeLocParser.createGenomeLoc(getHeader().getSequenceDictionary().getSequence(0).getSequenceIndex(), 0, Integer.MAX_VALUE);
@@ -312,6 +316,14 @@ public class SAMDataSource implements SimpleDataSource {
         return bound;
     }
 
+    /**
+     * Determines whether the BAM file is completely unsequenced.  Requires that the resource pool be initialized.
+     * @return True if the sequence dictionary is completely empty.  False otherwise.
+     */
+    private boolean isSequenceDictionaryEmpty() {
+        return getHeader().getSequenceDictionary().isEmpty();
+    }
+
     /**
      * Even though the iterator has seeked to the correct location, there may be multiple reads at that location,
      * and we may have given some of them out already.  Move the iterator to the correct location using the readsAtLastPos variable
diff --git a/java/src/org/broadinstitute/sting/gatk/iterators/MergingSamRecordIterator2.java b/java/src/org/broadinstitute/sting/gatk/iterators/MergingSamRecordIterator2.java
index 89e9aee85..931da02c1 100644
--- a/java/src/org/broadinstitute/sting/gatk/iterators/MergingSamRecordIterator2.java
+++ b/java/src/org/broadinstitute/sting/gatk/iterators/MergingSamRecordIterator2.java
@@ -154,7 +154,17 @@ public class MergingSamRecordIterator2 implements CloseableIterator<SAMRecord>,
         }
         final SAMRecordComparator comparator = getComparator();
         for (final SAMFileReader reader : samHeaderMerger.getReaders()) {
-            Iterator<SAMRecord> recordIter = reader.queryUnmapped();
+            Iterator<SAMRecord> recordIter = null;
+            if( reader.hasIndex() ) {
+                recordIter = reader.queryUnmapped();
+            }
+            else {
+                // HACK: Supporting completely unmapped BAM files is easy.  Let's do a quick check to make sure
+                //       these BAMs aren't partially indexed.
+                if( reader.getFileHeader().getSequenceDictionary().size() > 0 )
+                    throw new StingException("Partially mapped BAM files without indices are not supported");
+                recordIter = reader.iterator();
+            }
             final ComparableSamRecordIterator iterator = new ComparableSamRecordIterator(reader, recordIter, comparator);
             addIfNotEmpty(iterator);
         }