CSI index support for BAM files

src/main/java/htsjdk/samtools/BAMFileConstants.java

@@ -39,7 +39,19 @@ class BAMFileConstants {
 
     static final byte[] BAM_MAGIC = "BAM\1".getBytes();
     /**
-     * BAM index file magic number.
+     * BAM index file magic numbers.
+     * @deprecated prefer {@link BAMFileConstants#BAI_INDEX_MAGIC}
      */
+    @Deprecated
     static final byte[] BAM_INDEX_MAGIC = "BAI\1".getBytes();
+    static final byte[] BAI_INDEX_MAGIC = "BAI\1".getBytes();
+    static final byte[] CSI_INDEX_MAGIC = "CSI\1".getBytes();
+
+    /**
+     * CSI index related constants
+     */
+    static final int CSI_MAGIC_OFFSET = 0;
+    static final int CSI_MINSHIFT_OFFSET = 4;
+    static final int CSI_AUXDATA_OFFSET = 12;
+    static final int CSI_CHUNK_SIZE = 16;
 }

src/main/java/htsjdk/samtools/BAMFileReader.java

@@ -378,6 +378,9 @@ protected void enableIndexMemoryMapping(final boolean enabled) {
 
     @Override
     public SamReader.Type type() {
+        if (mIndexFile != null && getIndexType().equals(SamIndexes.CSI)) {
+            return SamReader.Type.BAM_CSI_TYPE;
+        }
         return SamReader.Type.BAM_TYPE;
     }
 
@@ -398,16 +401,41 @@ public BAMIndex getIndex() {
         if(!hasIndex())
             throw new SAMException("No index is available for this BAM file.");
         if(mIndex == null) {
-            if (mIndexFile != null)
-                mIndex = mEnableIndexCaching ? new CachingBAMFileIndex(mIndexFile, getFileHeader().getSequenceDictionary(), mEnableIndexMemoryMapping)
-                                             : new DiskBasedBAMFileIndex(mIndexFile, getFileHeader().getSequenceDictionary(), mEnableIndexMemoryMapping);
-            else
+            SamIndexes samIndex = getIndexType();
+            if (samIndex == null) {
                 mIndex = mEnableIndexCaching ? new CachingBAMFileIndex(mIndexStream, getFileHeader().getSequenceDictionary())
-                                             : new DiskBasedBAMFileIndex(mIndexStream, getFileHeader().getSequenceDictionary());
+                        : new DiskBasedBAMFileIndex(mIndexStream, getFileHeader().getSequenceDictionary());
+            } else if (samIndex.equals(SamIndexes.BAI)) {
+                    mIndex = mEnableIndexCaching ? new CachingBAMFileIndex(mIndexFile, getFileHeader().getSequenceDictionary(), mEnableIndexMemoryMapping)
+                            : new DiskBasedBAMFileIndex(mIndexFile, getFileHeader().getSequenceDictionary(), mEnableIndexMemoryMapping);
+            } else if (samIndex.equals(SamIndexes.CSI)) {
+                    mIndex = new CSIIndex(mIndexFile, mEnableIndexMemoryMapping, getFileHeader().getSequenceDictionary());
+            } else {
+                throw new SAMFormatException("Unsupported BAM index file: " + mIndexFile.getName());
+            }
         }
+
         return mIndex;
     }
 
+    /**
+     * Return the type of the BAM index, BAI or CSI.
+     * @return one of {@link SamIndexes#BAI} or {@link SamIndexes#CSI} or null
+     */
+    public SamIndexes getIndexType() {
+        if (mIndexFile != null) {
+            if (mIndexFile.getName().toLowerCase().endsWith(BAMIndex.BAI_INDEX_SUFFIX)) {
+                return SamIndexes.BAI;
+            } else if (mIndexFile.getName().toLowerCase().endsWith(BAMIndex.CSI_INDEX_SUFFIX)) {
+                return SamIndexes.CSI;
+            }
+
+            throw new SAMFormatException("Unknown BAM index file type: " + mIndexFile.getName());
+        }
+
+        return null;
+    }
+
     public void setEagerDecode(final boolean desired) { this.eagerDecode = desired; }
 
     @Override

src/main/java/htsjdk/samtools/BAMFileWriter.java

@@ -111,7 +111,7 @@ private BAMIndexer createBamIndex(final String pathURI) {
         try {
             final String indexFileBase = pathURI.endsWith(BamFileIoUtils.BAM_FILE_EXTENSION) ?
                     pathURI.substring(0, pathURI.lastIndexOf('.')) : pathURI;
-            final Path indexPath = IOUtil.getPath(indexFileBase + BAMIndex.BAMIndexSuffix);
+            final Path indexPath = IOUtil.getPath(indexFileBase + BAMIndex.BAI_INDEX_SUFFIX);
             if (Files.exists(indexPath)) {
                 if (!Files.isWritable(indexPath)) {
                     throw new SAMException("Not creating BAM index since unable to write index file " + indexPath.toUri());

src/main/java/htsjdk/samtools/BAMIndex.java

@@ -33,7 +33,13 @@
  */
 public interface BAMIndex extends Closeable {
 
-    public static final String BAMIndexSuffix = ".bai";
+    /**
+     * @deprecated prefer {@link BAMIndex#BAI_INDEX_SUFFIX} instead.
+     */
+    @Deprecated
+    String BAMIndexSuffix = ".bai";
+    String BAI_INDEX_SUFFIX = ".bai";
+    String CSI_INDEX_SUFFIX = ".csi";
 
     /**
      * Gets the compressed chunks which should be searched for the contents of records contained by the span
@@ -58,7 +64,7 @@ public interface BAMIndex extends Closeable {
      * @param reference the reference of interest
      * @return meta data for the reference
      */
-    public BAMIndexMetaData getMetaData(int reference);
+    BAMIndexMetaData getMetaData(int reference);
 
     /**
      * Close the index and release any associated resources.

src/main/java/htsjdk/samtools/BAMIndexMetaData.java

@@ -212,17 +212,22 @@ long getLastOffset() {
     }
 
     /**
-     * Prints meta-data statistics from BAM index (.bai) file
+     * Prints meta-data statistics from BAM index (.bai or .csi) file
      * Statistics include count of aligned and unaligned reads for each reference sequence
      * and a count of all records with no start coordinate
      */
     static public void printIndexStats(final File inputBamFile) {
         try {
             final BAMFileReader bam = new BAMFileReader(inputBamFile, null, false, false, ValidationStringency.SILENT, new DefaultSAMRecordFactory());
-            if (!bam.hasIndex()) {
+            if (!bam.hasIndex() || bam.getIndexType() == null) {
                 throw new SAMException("No index for bam file " + inputBamFile);
             }
+
             BAMIndexMetaData[] data = getIndexStats(bam);
+            if (data == null) {
+                throw new SAMException("Exception in getting index statistics");
+            }
+
             // read through all the bins of every reference.
             int nRefs = bam.getFileHeader().getSequenceDictionary().size();
             for (int i = 0; i < nRefs; i++) {
@@ -245,7 +250,7 @@ static public void printIndexStats(final File inputBamFile) {
     }
 
     /**
-     * Prints meta-data statistics from BAM index (.bai) file
+     * Prints meta-data statistics from BAM index (.bai or .csi) file
      * Statistics include count of aligned and unaligned reads for each reference sequence
      * and a count of all records with no start coordinate
      */

src/main/java/htsjdk/samtools/BamFileIoUtils.java

@@ -126,7 +126,7 @@ public static void gatherWithBlockCopying(final List<File> bams, final File outp
             if (createMd5) out = new Md5CalculatingOutputStream(out, new File(output.getAbsolutePath() + ".md5"));
             File indexFile = null;
             if (createIndex) {
-                indexFile = new File(output.getParentFile(), IOUtil.basename(output) + BAMIndex.BAMIndexSuffix);
+                indexFile = new File(output.getParentFile(), IOUtil.basename(output) + BAMIndex.BAI_INDEX_SUFFIX);
                 out = new StreamInflatingIndexingOutputStream(out, indexFile);
             }
 
@@ -161,7 +161,7 @@ private static OutputStream buildOutputStream(final File outputFile, final boole
             outputStream = new Md5CalculatingOutputStream(outputStream, new File(outputFile.getAbsolutePath() + ".md5"));
         }
         if (createIndex) {
-            outputStream = new StreamInflatingIndexingOutputStream(outputStream, new File(outputFile.getParentFile(), IOUtil.basename(outputFile) + BAMIndex.BAMIndexSuffix));
+            outputStream = new StreamInflatingIndexingOutputStream(outputStream, new File(outputFile.getParentFile(), IOUtil.basename(outputFile) + BAMIndex.BAI_INDEX_SUFFIX));
         }
         return outputStream;
     }