htsjdk.samtools.util.BinaryCodec Java Examples

/**
 * Writes out a SAMRecord in Maq fastq format
 *
 * @param codec the code to write to
 * @param rec   the SAMRecord to write
 */
private void writeFastqRecord(final BinaryCodec codec, final SAMRecord rec) {

    // Trim the run barcode off the read name
    String readName = rec.getReadName();
    if (namePrefix != null && readName.startsWith(namePrefix)) {
        readName = readName.substring(nameTrim);
    }
    // Writes the length of the read name and then the name (null-terminated)
    codec.writeString(readName, true, true);

    final char[] seqs = rec.getReadString().toCharArray();
    final char[] quals = rec.getBaseQualityString().toCharArray();

    int retainedLength = seqs.length;
    if (clipAdapters){
        // adjust to a shorter length iff clipping tag exists
        Integer trimPoint = rec.getIntegerAttribute(ReservedTagConstants.XT);
        if (trimPoint != null) {
            ValidationUtils.validateArg(rec.getReadLength() == seqs.length, () -> "length of read and seqs differ. Found " + rec.getReadLength() + " and '" + seqs.length + ".");

            retainedLength = Math.min(seqs.length, Math.max(SEED_REGION_LENGTH, trimPoint -1));
        }
    }

    // Write the length of the sequence
    codec.writeInt(basesToWrite != null ? basesToWrite : seqs.length);

    // Calculate and write the sequence and qualities
    final byte[] seqsAndQuals = encodeSeqsAndQuals(seqs, quals, retainedLength);
    codec.writeBytes(seqsAndQuals);
}
 

/**
 * @param nRef Number of reference sequences. If zero is passed then header is not written.
 * This is useful in sharded writing as we only want the header written for the first shard.
 * 
 * @param output BAM index output stream.  This stream will be closed when BinaryBAMIndexWriter.close() is called.
 */
public BinaryBAMShardIndexWriter(final int nRef, final OutputStream output) {
    try {
        codec = new BinaryCodec(output);
        if (nRef > 0) {
          writeHeader(nRef);
        }
    } catch (final Exception e) {
        throw new SAMException("Exception opening output stream", e);
    }
}
 

/**
 * Private helper method for {@link #convertHeaderlessHadoopBamShardToBam} that takes a SAMFileHeader and writes it
 * to the provided `OutputStream`, correctly encoded for the BAM format and preceded by the BAM magic bytes.
 *
 * @param samFileHeader SAM header to write
 * @param outputStream stream to write the SAM header to
 */
private static void writeBAMHeaderToStream( final SAMFileHeader samFileHeader, final OutputStream outputStream ) {
    final BlockCompressedOutputStream blockCompressedOutputStream = new BlockCompressedOutputStream(outputStream, (File)null);
    final BinaryCodec outputBinaryCodec = new BinaryCodec(new DataOutputStream(blockCompressedOutputStream));

    final String headerString;
    final Writer stringWriter = new StringWriter();
    new SAMTextHeaderCodec().encode(stringWriter, samFileHeader, true);
    headerString = stringWriter.toString();

    outputBinaryCodec.writeBytes(ReadUtils.BAM_MAGIC);

    // calculate and write the length of the SAM file header text and the header text
    outputBinaryCodec.writeString(headerString, true, false);

    // write the sequences binarily.  This is redundant with the text header
    outputBinaryCodec.writeInt(samFileHeader.getSequenceDictionary().size());
    for (final SAMSequenceRecord sequenceRecord: samFileHeader.getSequenceDictionary().getSequences()) {
        outputBinaryCodec.writeString(sequenceRecord.getSequenceName(), true, true);
        outputBinaryCodec.writeInt(sequenceRecord.getSequenceLength());
    }

    try {
        blockCompressedOutputStream.flush();
    } catch (final IOException ioe) {
        throw new RuntimeIOException(ioe);
    }
}
 

public Bam_OBA_Supplier(BufferedInputStream is) {
	this.is = is;
	codec = new BinaryCodec();
	codec.setInputStream(is);
	baos = new ByteArrayOutputStream();
	refId = Integer.MIN_VALUE;
	recordCounter = 0;
}
 

private static SAMSequenceRecord readSequenceRecord(final BinaryCodec stream, final String source) {
	final int nameLength = stream.readInt();
	if (nameLength <= 1) {
		throw new SAMFormatException("Invalid BAM file header: missing sequence name in file " + source);
	}
	final String sequenceName = stream.readString(nameLength - 1);
	// Skip the null terminator
	stream.readByte();
	final int sequenceLength = stream.readInt();
	return new SAMSequenceRecord(SAMSequenceRecord.truncateSequenceName(sequenceName), sequenceLength);
}
 

static void writeHeader(final BinaryCodec outputBinaryCodec, final SAMFileHeader samFileHeader,
		final String headerText) {
	outputBinaryCodec.writeBytes("BAM\1".getBytes());

	// calculate and write the length of the SAM file header text and the
	// header text
	outputBinaryCodec.writeString(headerText, true, false);

	// write the sequences binarily. This is redundant with the text header
	outputBinaryCodec.writeInt(samFileHeader.getSequenceDictionary().size());
	for (final SAMSequenceRecord sequenceRecord : samFileHeader.getSequenceDictionary().getSequences()) {
		outputBinaryCodec.writeString(sequenceRecord.getSequenceName(), true, true);
		outputBinaryCodec.writeInt(sequenceRecord.getSequenceLength());
	}
}
 

/**
 * Writes a header to a BAM file. Might need to regenerate the String
 * version of the header, if one already has both the samFileHeader and the
 * String, use the version of this method which takes both.
 */
static void writeHeader(final BinaryCodec outputBinaryCodec, final SAMFileHeader samFileHeader) {
	// Do not use SAMFileHeader.getTextHeader() as it is not updated when
	// changes to the underlying object are made
	final String headerString;
	final Writer stringWriter = new StringWriter();
	new SAMTextHeaderCodec().encode(stringWriter, samFileHeader, true);
	headerString = stringWriter.toString();

	writeHeader(outputBinaryCodec, samFileHeader, headerString);
}
 

protected static void writeHeader(final OutputStream outputStream, final SAMFileHeader samFileHeader) {
	final BlockCompressedOutputStream blockCompressedOutputStream = new BlockCompressedOutputStream(outputStream,
			null);
	final BinaryCodec outputBinaryCodec = new BinaryCodec(new DataOutputStream(blockCompressedOutputStream));
	writeHeader(outputBinaryCodec, samFileHeader);
	try {
		blockCompressedOutputStream.flush();
	} catch (final IOException ioe) {
		throw new RuntimeIOException(ioe);
	}
}
 

static SAMFileHeader readHeader(final BinaryCodec stream, final ValidationStringency validationStringency,
		final String source) throws IOException {

	final byte[] buffer = new byte[4];
	stream.readBytes(buffer);
	if (!Arrays.equals(buffer, "BAM\1".getBytes())) {
		throw new IOException("Invalid BAM file header");
	}

	final int headerTextLength = stream.readInt();
	final String textHeader = stream.readString(headerTextLength);
	final SAMTextHeaderCodec headerCodec = new SAMTextHeaderCodec();
	headerCodec.setValidationStringency(validationStringency);
	final SAMFileHeader samFileHeader = headerCodec.decode(new StringLineReader(textHeader), source);

	final int sequenceCount = stream.readInt();
	if (samFileHeader.getSequenceDictionary().size() > 0) {
		// It is allowed to have binary sequences but no text sequences, so
		// only validate if both are present
		if (sequenceCount != samFileHeader.getSequenceDictionary().size()) {
			throw new SAMFormatException("Number of sequences in text header ("
					+ samFileHeader.getSequenceDictionary().size() + ") != number of sequences in binary header ("
					+ sequenceCount + ") for file " + source);
		}
		for (int i = 0; i < sequenceCount; i++) {
			final SAMSequenceRecord binarySequenceRecord = readSequenceRecord(stream, source);
			final SAMSequenceRecord sequenceRecord = samFileHeader.getSequence(i);
			if (!sequenceRecord.getSequenceName().equals(binarySequenceRecord.getSequenceName())) {
				throw new SAMFormatException("For sequence " + i
						+ ", text and binary have different names in file " + source);
			}
			if (sequenceRecord.getSequenceLength() != binarySequenceRecord.getSequenceLength()) {
				throw new SAMFormatException("For sequence " + i
						+ ", text and binary have different lengths in file " + source);
			}
		}
	} else {
		// If only binary sequences are present, copy them into
		// samFileHeader
		final List<SAMSequenceRecord> sequences = new ArrayList<SAMSequenceRecord>(sequenceCount);
		for (int i = 0; i < sequenceCount; i++) {
			sequences.add(readSequenceRecord(stream, source));
		}
		samFileHeader.setSequenceDictionary(new SAMSequenceDictionary(sequences));
	}

	return samFileHeader;
}