htsjdk.samtools.SAMFileReader Java Examples

public static void sortSamBam(File samBamFile, File sortedBamFile) {
	
	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
	SAMFileReader reader = new SAMFileReader(IOUtil.openFileForReading(samBamFile));
	SAMFileWriter writer = null;
	try {
		
		reader.getFileHeader().setSortOrder(SAMFileHeader.SortOrder.coordinate);
		writer = new SAMFileWriterFactory().makeBAMWriter(reader.getFileHeader(), false, sortedBamFile);
		Iterator<SAMRecord> iterator = reader.iterator();
		while (iterator.hasNext()) {
			writer.addAlignment(iterator.next());
		}
		
	} finally {
		closeIfPossible(reader);
		closeIfPossible(writer);
	}
}
 

public void indexBam(File bamFile, File baiFile) {
	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
       final SamReader bam;

           // input from a normal file
           IOUtil.assertFileIsReadable(bamFile);
           bam = SamReaderFactory.makeDefault().referenceSequence(null)
                   .enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS)
                   .open(bamFile);

       if (bam.type() != SamReader.Type.BAM_TYPE) {
           throw new SAMException("Input file must be bam file, not sam file.");
       }

       if (!bam.getFileHeader().getSortOrder().equals(SAMFileHeader.SortOrder.coordinate)) {
           throw new SAMException("Input bam file must be sorted by coordinate");
       }

       BAMIndexer.createIndex(bam, baiFile);

       CloserUtil.close(bam);
}
 

public String printSamBam(InputStream samBamStream, int maxRecords) throws IOException {
	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
	SAMFileReader in = new SAMFileReader(samBamStream);
	SAMFileHeader header = in.getFileHeader();
	ByteArrayOutputStream buffer = new ByteArrayOutputStream();
	SAMFileWriter out = new SAMFileWriterFactory().makeSAMWriter(header, true, buffer);
	int i = 0;
	try {
		for (final SAMRecord rec : in) {
			if (i > maxRecords) {
				break;
			}
			out.addAlignment(rec);
			i++;
		}
	} finally {
		closeIfPossible(out);
	}

	if (i > maxRecords) {
		buffer.write("SAM/BAM too long for viewing, truncated here!\n".getBytes());
	}
	
	return buffer.toString();
}
 

private SAMRecord buildSAMRecord(String seqName, String line) {
	ByteArrayOutputStream baos = new ByteArrayOutputStream();
	try {
		baos.write("@HD\tVN:1.0\tGO:none SO:coordinate\n".getBytes());
		baos.write(("@SQ\tSN:" + seqName + "\tLN:247249719\n").getBytes());
		baos.write(line.replaceAll("\\s+", "\t").getBytes());
		baos.close();
	} catch (IOException e) {
		throw new RuntimeException(e);
	}

	ByteArrayInputStream bais = new ByteArrayInputStream(baos.toByteArray());
	SAMFileReader r = new SAMFileReader(bais);
	try {
		return r.iterator().next();
	} finally {
		r.close();
	}
}
 

/**
  * Generally we would like to have both data and index files,
  * because otherwise we cannot access random locations.
  * 
  * @param data
  * @param index
  * @throws URISyntaxException
  * @throws IOException 
  */
 public BamDataSource(DataUrl data, DataUrl index) throws URISyntaxException, IOException {
     super(data);

 	// BAMFileReader emits useless warning to System.err that can't be turned off,
 	// so we direct it to other stream and discard. 
 	PrintStream originalErr = System.err;
 	System.setErr(new PrintStream(new ByteArrayOutputStream()));
 	
 	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
 	this.reader = SamBamUtils.getSAMReader(data.getUrl(), index.getUrl());

 	LinkedList<String> chrList = new LinkedList<>();
 	
 	// Iterate chromosomes to check naming convention
 	for (SAMSequenceRecord sequenceRecord : this.reader.getFileHeader().getSequenceDictionary().getSequences()) {
 		
 		String name = sequenceRecord.getSequenceName();
 		
chrList.add(name);
 	}
 	
 	// Create unnormaliser for this naming convention
 	// Look only at the first sequence (assume all have the same convention)
 	this.chromosomeNameUnnormaliser = new ChromosomeNameUnnormaliser(chrList);      
     
     // Restore System.err
     System.setErr(originalErr);
 }
 

public void normaliseBam(File bamFile, File normalisedBamFile) {

		// Read in a BAM file and its header
		SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
		SAMFileReader reader = new SAMFileReader(IOUtil.openFileForReading(bamFile));
		SAMFileWriter writer = null;
		try {
			SAMFileHeader normalisedHeader = reader.getFileHeader();

			// Alter the chromosome names in header's SAMSequenceDictionary
			SAMSequenceDictionary normalisedDictionary = new SAMSequenceDictionary();
			for (SAMSequenceRecord sequenceRecord : normalisedHeader.getSequenceDictionary().getSequences()) {

				// Normalise chromosome
				String sequenceName = chromosomeNormaliser.normaliseChromosome(sequenceRecord.getSequenceName());
				normalisedDictionary.addSequence(new SAMSequenceRecord(sequenceName, sequenceRecord.getSequenceLength()));
			}
			normalisedHeader.setSequenceDictionary(normalisedDictionary);

			// Write new BAM file with normalised chromosome names
			writer = new SAMFileWriterFactory().makeBAMWriter(normalisedHeader, true, normalisedBamFile);
			for (final SAMRecord rec : reader) {
				rec.setHeader(normalisedHeader);
				writer.addAlignment(rec);
			}
			
		} finally {
			closeIfPossible(reader);
			closeIfPossible(writer);
		}
	}
 

public static void closeIfPossible(SAMFileReader reader) {
	if (reader != null) {
		try {
			reader.close();
		} catch (Exception e) {
			// Ignore
		}
	}
}
 

/**
 * Read the data and create a list of lengths
 */
private void read(){
	int counter = 0;
	SAMFileReader reader = new SAMFileReader(bam,index);
	ArrayList<Double> temp = new ArrayList<Double>();
	for (int i = 0; i < genome.size();i++){
		String chr = genome.get(i).getChrom();
		int start = genome.get(i).getStart();
		int stop = genome.get(i).getStop();
		CloseableIterator<SAMRecord> iter = reader.query(chr,start,stop,false);
		while (iter.hasNext()){
			SAMRecord record = null;
			try{
				record = iter.next();
			}
			catch(SAMFormatException ex){
				System.out.println("SAM Record is problematic. Has mapQ != 0 for unmapped read. Will continue anyway");
			}
			if(record != null){
			if(!record.getReadUnmappedFlag() && !record.getMateUnmappedFlag() && record.getFirstOfPairFlag()) {
				if (record.getMappingQuality() >= minQ){
					
					if (Math.abs(record.getInferredInsertSize()) > 100 && Math.abs(record.getInferredInsertSize())
							< 1000){
						counter+=1;
						temp.add((double)Math.abs(record.getInferredInsertSize()));
					}
				}
			}
			}
		}
		iter.close();
	}
	reader.close();
	lengths = new double[counter];
	for (int i = 0;i < temp.size();i++){
		if (temp.get(i) > 100){
			lengths[i] = temp.get(i);
		}
	}
	
}
 

@Override
public void doRun() throws IOException {
	super.doRun();

	fo.close();

	if (fo.empty)
		return;

	log.info("Sorting overflow BAM: ", fo.file.length());
	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
	SAMFileReader r = new SAMFileReader(fo.file);
	SAMRecordIterator iterator = r.iterator();
	if (!iterator.hasNext()) {
		r.close();
		fo.file.delete();
		return;
	}

	SAMRecord r1 = iterator.next();
	SAMRecord r2 = null;
	counter = multiFastqOutputter.getCounter();
	log.info("Counter=" + counter);
	while (!brokenPipe.get() && iterator.hasNext()) {
		r2 = iterator.next();
		if (r1.getReadName().equals(r2.getReadName())) {
			print(r1, r2);
			counter++;
			r1 = null;
			if (!iterator.hasNext())
				break;
			r1 = iterator.next();
			r2 = null;
		} else {
			print(r1, 0);
			r1 = r2;
			r2 = null;
			counter++;
		}
	}
	if (r1 != null)
		print(r1, 0);
	r.close();
	fo.file.delete();
}
 

@Test
public void test() throws IOException {
	byte[] buf = new byte[1024];
	BAMRecordView view = new BAMRecordView(buf);
	view.setRefID(0);
	view.setAlignmentStart(77);
	view.setMappingScore(44);
	view.setIndexBin(99);
	view.setFlags(555);
	view.setMateRefID(0);
	view.setMateAlStart(78);
	view.setInsertSize(133);

	view.setReadName("name1");
	view.setCigar(TextCigarCodec.decode("10M"));
	view.setBases("AAAAAAAAAA".getBytes());
	view.setQualityScores("BBBBBBBBBB".getBytes());

	int id = 'A' << 16 | 'M' << 8 | 'A';
	view.addTag(id, "Q".getBytes(), 0, 1);

	int len = view.finish();

	System.out.println(Arrays.toString(Arrays.copyOf(buf, len)));

	ByteArrayOutputStream baos = new ByteArrayOutputStream();

	SAMFileHeader header = new SAMFileHeader();
	header.addSequence(new SAMSequenceRecord("14", 14));

	ByteArrayOutputStream baos2 = new ByteArrayOutputStream();
	SAMFileWriter writer = new SAMFileWriterFactory().makeBAMWriter(header, true, baos2);
	SAMRecord record = new SAMRecord(header);
	record.setReferenceIndex(0);
	record.setAlignmentStart(1);
	record.setCigarString("10M");
	record.setFlags(555);
	record.setMappingQuality(44);
	record.setMateReferenceIndex(0);
	record.setMateAlignmentStart(0);
	record.setInferredInsertSize(133);
	record.setReadName("name1");
	record.setReadBases("AAAAAAAAAA".getBytes());
	record.setBaseQualities("BBBBBBBBBB".getBytes());
	record.setAttribute("AM", 'Q');

	System.out.println("BAMFileWriter.addAlignment():");
	writer.addAlignment(record);
	System.out.println(".");
	writer.close();

	System.out.println("------------------------------------------");
	System.out.println();
	System.out.println(new String(baos2.toByteArray()));
	System.out.println();

	SAMFileReader.setDefaultValidationStringency(ValidationStringency.SILENT);
	SAMFileReader reader2 = new SAMFileReader(new ByteArrayInputStream(baos2.toByteArray()));
	SAMRecordIterator iterator = reader2.iterator();
	while (iterator.hasNext()) {
		record = iterator.next();
		System.out.println(record.getSAMString());
	}
	System.out.println("------------------------------------------");

	BlockCompressedOutputStream bcos = new BlockCompressedOutputStream(baos, null);
	bcos.write("BAM\1".getBytes());
	bcos.write(toByteArray(header));
	CramInt.writeInt32(header.getSequenceDictionary().size(), bcos);
	for (final SAMSequenceRecord sequenceRecord : header.getSequenceDictionary().getSequences()) {
		byte[] bytes = sequenceRecord.getSequenceName().getBytes();
		CramInt.writeInt32(bytes.length + 1, bcos);
		bcos.write(sequenceRecord.getSequenceName().getBytes());
		bcos.write(0);
		CramInt.writeInt32(sequenceRecord.getSequenceLength(), bcos);
	}
	bcos.write(buf, 0, len);
	bcos.close();

	System.out.println(new String(baos.toByteArray()));

	SAMFileReader reader = new SAMFileReader(new ByteArrayInputStream(baos.toByteArray()));
	iterator = reader.iterator();
	while (iterator.hasNext()) {
		record = iterator.next();
		System.out.println(record.getSAMString());
	}
	reader.close();

}
 

public static List<String> readChromosomeNames(URL bam, URL index) throws FileNotFoundException, URISyntaxException {
								
		SAMFileReader reader = getSAMReader(bam, index);
		
		LinkedList<String> chromosomes = new LinkedList<String>();
		for (SAMSequenceRecord record : reader.getFileHeader().getSequenceDictionary().getSequences()) {
			chromosomes.add(record.getSequenceName());
		}
		
		closeIfPossible(reader);
		
		return chromosomes;		
}