AddDummyAlignmentInformation.java

AddDummyAlignmentInformation.java

package picard.sam;

import htsjdk.samtools.SAMFileHeader;
import htsjdk.samtools.SAMFileWriter;
import htsjdk.samtools.SAMFileWriterFactory;
import htsjdk.samtools.SAMRecord;
import htsjdk.samtools.SAMSequenceRecord;
import htsjdk.samtools.SamReader;
import htsjdk.samtools.SamReaderFactory;
import htsjdk.samtools.util.CloserUtil;
import htsjdk.samtools.util.Log;
import htsjdk.samtools.util.ProgressLogger;
import picard.PicardException;
import picard.cmdline.CommandLineProgram;
import picard.cmdline.CommandLineProgramProperties;
import picard.cmdline.Option;
import picard.cmdline.StandardOptionDefinitions;
import picard.cmdline.programgroups.SamOrBam;

import java.io.File;

@CommandLineProgramProperties(
        usage = "Adds dummy alignment information to a query-sorted unmapped SAM or BAM.",
        usageShort = "Adds dummy alignment information to a query-sorted unmapped SAM or BAM.",
        programGroup = SamOrBam.class
)
public class AddDummyAlignmentInformation extends CommandLineProgram {

    private static final Log log = Log.getInstance(AddDummyAlignmentInformation.class);

    @Option(shortName = StandardOptionDefinitions.INPUT_SHORT_NAME, doc = "The SAM or BAM file or GA4GH url to view.")
    public File INPUT;

    @Option(shortName = StandardOptionDefinitions.INPUT_SHORT_NAME, doc = "The BAM file to write.")
    public File OUTPUT;

    /** Stock main method. */
    public static void main(final String[] argv) {
        System.exit(new AddDummyAlignmentInformation().instanceMain(argv));
    }

    protected int doWork() {
        final SamReader reader = SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(INPUT);

        if (reader.getFileHeader().getSortOrder() != SAMFileHeader.SortOrder.queryname) {
            throw new PicardException("INPUT must be in queryname order");
        }

        final SAMFileHeader header = reader.getFileHeader();
        header.addSequence(new SAMSequenceRecord("DUMMY", Integer.MAX_VALUE));
        final SAMFileWriter writer = new SAMFileWriterFactory().setCreateIndex(true).makeBAMWriter(header, false, OUTPUT);

        int recordIndex = 1;
        SAMRecord prevRecord = null;
        final ProgressLogger progress = new ProgressLogger(log);
        for (final SAMRecord record : reader) {
            // read
            record.setReadUnmappedFlag(true);
            record.setReferenceIndex(0);
            record.setAlignmentStart(recordIndex);
            // mate info
            record.setMateUnmappedFlag(true);
            record.setMateReferenceIndex(0);
            record.setMateAlignmentStart(recordIndex);

            writer.addAlignment(record);
            progress.record(record);

            if (recordIndex == Integer.MAX_VALUE) {
                throw new PicardException(String.format("Too many records (>%d)", Integer.MAX_VALUE));
            }

            if (null == prevRecord || record.getReadName().equals(prevRecord.getReadName())) {
                prevRecord = record;
                recordIndex++;
            }
        }
        log.info(String.format("Wrote %d records.", recordIndex-1));

        writer.close();
        CloserUtil.close(reader);

        return 0;
    }
}