JAL-2909 add sequence feature for insertions

[jalview.git] / src / jalview / io / BamFile.java

diff --git a/src/jalview/io/BamFile.java b/src/jalview/io/BamFile.java
index 26c14a2..10e2d0e 100644 (file)
--- a/src/jalview/io/BamFile.java
+++ b/src/jalview/io/BamFile.java
@@ -26,18 +26,33 @@ import jalview.datamodel.SequenceI;
 
 import java.io.File;
 import java.io.IOException;
+import java.util.ArrayList;
+import java.util.List;
 import java.util.SortedMap;
 
 import htsjdk.samtools.SAMRecord;
 import htsjdk.samtools.SAMRecordIterator;
+import htsjdk.samtools.SAMSequenceRecord;
 import htsjdk.samtools.SamReader;
 import htsjdk.samtools.SamReaderFactory;
 import htsjdk.samtools.ValidationStringency;
 
 public class BamFile extends AlignFile
 {
+  // SAM/BAM file reader
+  private SamReader fileReader;
 
-  SamReader fileReader;
+  // start position to read from
+  private int start = -1;
+
+  // end position to read to
+  private int end = -1;
+
+  // chromosome/contig to read
+  private String chromosome = "";
+
+  // first position in alignment
+  private int alignmentStart = -1;
 
   /**
    * Creates a new BamFile object.
@@ -50,17 +65,16 @@ public class BamFile extends AlignFile
    * Creates a new BamFile object.
    * 
    * @param inFile
-   *          DOCUMENT ME!
+   *          Name of file to read
    * @param sourceType
-   *          DOCUMENT ME!
+   *          Whether data source is file, url or other type of data source
    * 
    * @throws IOException
-   *           DOCUMENT ME!
    */
   public BamFile(String inFile, DataSourceType sourceType)
           throws IOException
   {
-    super(inFile, sourceType);
+    super(true, inFile, sourceType);
     final SamReaderFactory factory = SamReaderFactory.makeDefault()
             .enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS,
                     SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS)
@@ -68,8 +82,16 @@ public class BamFile extends AlignFile
     fileReader = factory.open(new File(inFile));
   }
 
+  /**
+   * Creates a new BamFile object
+   * 
+   * @param source
+   *          wrapper for datasource
+   * @throws IOException
+   */
   public BamFile(FileParse source) throws IOException
   {
+    super(true, source);
     final SamReaderFactory factory = SamReaderFactory.makeDefault()
             .enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS,
                     SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS)
@@ -77,7 +99,6 @@ public class BamFile extends AlignFile
 
     // File-based bam
     fileReader = factory.open(source.inFile);
-    parse();
   }
   
   @Override
@@ -88,33 +109,73 @@ public class BamFile extends AlignFile
   }
 
   @Override
-  public void parse() throws IOException
+  public void parse()
   {
-    SAMRecordIterator it = fileReader.iterator();
-    CigarParser parser = new CigarParser('-');
-    SortedMap<Integer, Integer> insertions = parser.getInsertions(it);
-    it.close();
-
-    it = fileReader.iterator();
-    while (it.hasNext())
+    // only actually parse if params are set
+    if (chromosome != null && chromosome != "")
     {
-      SAMRecord rec = it.next();
+      SAMRecordIterator it = fileReader.query(chromosome, start, end,
+              false);
+      CigarParser parser = new CigarParser('-');
+      SortedMap<Integer, Integer> insertions = parser.getInsertions(it);
+      it.close();
+
+      it = fileReader.query(chromosome, start, end, false);
+      while (it.hasNext())
+      {
+        SAMRecord rec = it.next();
 
-      // make dataset sequence: start at 1, end at read length
-      SequenceI seq = new Sequence(rec.getReadName(),
-              rec.getReadString().toLowerCase());
-      seq.setStart(1);
-      seq.setEnd(rec.getReadLength());
+        // set the alignment start to be start of first read (we assume reads
+        // are sorted)
+        if (alignmentStart == -1)
+        {
+          alignmentStart = rec.getAlignmentStart();
+        }
 
-      String newRead = parser.parseCigarToSequence(rec, insertions);
+        // make dataset sequence: start at 1, end at read length
+        SequenceI seq = new Sequence(rec.getReadName(),
+                rec.getReadString().toLowerCase());
+        seq.setStart(1);
+        seq.setEnd(rec.getReadLength());
 
-      // make alignment sequences
-      SequenceI alsq = seq.deriveSequence();
-      alsq.setSequence(newRead);
+        String newRead = parser.parseCigarToSequence(rec, insertions,
+                alignmentStart, seq);
 
-      // set start relative to soft clip; assume end is set by Sequence code
-      alsq.setStart(rec.getStart() - rec.getUnclippedStart() + 1);
-      seqs.add(alsq);
+        // make alignment sequences
+        SequenceI alsq = seq.deriveSequence();
+        alsq.setSequence(newRead);
+
+        // set start relative to soft clip; assume end is set by Sequence code
+        alsq.setStart(rec.getStart() - rec.getUnclippedStart() + 1);
+        seqs.add(alsq);
+      }
     }
   }
+
+  /**
+   * Get the list of chromosomes or contigs from the file (listed in SQ entries
+   * in BAM file header)
+   * 
+   * @return array of chromosome/contig strings
+   */
+  @Override
+  public Object[] preprocess()
+  {
+    List<SAMSequenceRecord> refSeqs = fileReader.getFileHeader()
+            .getSequenceDictionary().getSequences();
+    List<String> chrs = new ArrayList<>();
+
+    for (SAMSequenceRecord ref : refSeqs)
+    {
+      chrs.add(ref.getSequenceName());
+    }
+    return chrs.toArray();
+  }
+
+  public void setOptions(String chr, int s, int e)
+  {
+    chromosome = chr;
+    start = s;
+    end = e;
+  }
 }