htsjdk.samtools.SAMValidationError Java Examples

@Test(dataProvider = "testCleanSamDataProvider")
public void testCleanSam(final String samFile, final String expectedCigar) throws IOException {
    final File cleanedFile = File.createTempFile(samFile + ".", ".sam");
    cleanedFile.deleteOnExit();
    final String[] args = new String[]{
            "INPUT=" + new File(TEST_DATA_DIR, samFile).getAbsolutePath(),
            "OUTPUT=" + cleanedFile.getAbsolutePath()
    };
    Assert.assertEquals(runPicardCommandLine(args), 0);

    final SamFileValidator validator = new SamFileValidator(new PrintWriter(System.out), 8000);
    validator.setIgnoreWarnings(true);
    validator.setVerbose(true, 1000);
    validator.setErrorsToIgnore(Arrays.asList(SAMValidationError.Type.MISSING_READ_GROUP));
    SamReader samReader = SamReaderFactory.makeDefault().validationStringency(ValidationStringency.LENIENT).open(cleanedFile);
    final SAMRecord rec = samReader.iterator().next();
    samReader.close();
    Assert.assertEquals(rec.getCigarString(), expectedCigar);
    samReader = SamReaderFactory.makeDefault().validationStringency(ValidationStringency.LENIENT).open(cleanedFile);
    final boolean validated = validator.validateSamFileVerbose(samReader, null);
    samReader.close();
    Assert.assertTrue(validated, "ValidateSamFile failed");
}
 

protected void test() {
    try {
        final SamFileValidator validator = new SamFileValidator(new PrintWriter(System.out), 8000);

        // Validate it has the expected cigar
        validator.setIgnoreWarnings(true);
        validator.setVerbose(true, 1000);
        validator.setErrorsToIgnore(Arrays.asList(SAMValidationError.Type.MISSING_READ_GROUP));
        SamReaderFactory factory = SamReaderFactory.makeDefault().validationStringency(ValidationStringency.LENIENT);
        SamReader samReader = factory.open(getOutput());
        final SAMRecordIterator iterator = samReader.iterator();
        while (iterator.hasNext()) {
            final SAMRecord rec = iterator.next();
            Assert.assertEquals(rec.getCigarString(), expectedCigar);
            if (SAMUtils.hasMateCigar(rec)) {
                Assert.assertEquals(SAMUtils.getMateCigarString(rec), expectedCigar);
            }
        }
        CloserUtil.close(samReader);

        // Run validation on the output file
        samReader = factory.open(getOutput());
        final boolean validated = validator.validateSamFileVerbose(samReader, null);
        CloserUtil.close(samReader);

        Assert.assertTrue(validated, "ValidateSamFile failed");
    } finally {
        IOUtil.recursiveDelete(getOutputDir().toPath());
    }
}
 

protected int doWork() {
    IOUtil.assertFileIsReadable(INPUT);
    final SamReader reader = SamReaderFactory.makeDefault().referenceSequence(REFERENCE_SEQUENCE).open(INPUT);
    final Map<String, SAMRecord> firstSeenMates = new HashMap<>();
    final FastqWriterFactory factory = new FastqWriterFactory();
    factory.setCreateMd5(CREATE_MD5_FILE);

    initializeAdditionalWriters();
    final Map<SAMReadGroupRecord, FastqWriters> writers = generateWriters(reader.getFileHeader().getReadGroups(),
            factory);
    final Map<SAMReadGroupRecord, List<FastqWriter>> additionalWriters = generateAdditionalWriters(reader.getFileHeader().getReadGroups(), factory);
    if (writers.isEmpty()) {
        final String msgBase = INPUT + " does not contain Read Groups";
        final String msg = OUTPUT_PER_RG ? msgBase + ", consider not using the OUTPUT_PER_RG option" : msgBase;
        throw new PicardException(msg);
    }

    final ProgressLogger progress = new ProgressLogger(log);

    for (final SAMRecord currentRecord : reader) {
        handleRecord(currentRecord, writers, additionalWriters, firstSeenMates);
        progress.record(currentRecord);
    }

    CloserUtil.close(reader);

    // Close all the fastq writers being careful to close each one only once!
    for (final FastqWriters writerMapping : new HashSet<>(writers.values())) {
        writerMapping.closeAll();
    }

    // close all `additionalWriters` only once
    final Set<FastqWriter> additionalWriterSet = new HashSet<>();
    additionalWriters.values().forEach(additionalWriterSet::addAll);
    for (final FastqWriter fastqWriter : additionalWriterSet) {
        fastqWriter.close();
    }

    if (!firstSeenMates.isEmpty()) {
        SAMUtils.processValidationError(new SAMValidationError(SAMValidationError.Type.MATE_NOT_FOUND,
                "Found " + firstSeenMates.size() + " unpaired mates", null), VALIDATION_STRINGENCY);
    }

    return 0;
}
 

public void setIgnoreError(final Collection<SAMValidationError.Type> errorsToIgnore){
    ignoreErrors = new ArrayList<>(errorsToIgnore);
}
 

public static void main(String[] args) throws IOException, IllegalArgumentException, IllegalAccessException {
	Params params = new Params();
	JCommander jc = new JCommander(params);
	try {
		jc.parse(args);
	} catch (Exception e) {
		System.out.println("Failed to parse parameteres, detailed message below: ");
		System.out.println(e.getMessage());
		System.out.println();
		System.out.println("See usage: -h");
		System.exit(1);
	}

	if (args.length == 0 || params.help) {
		printUsage(jc);
		System.exit(1);
	}

	if (params.reference == null) {
		System.out.println("A reference fasta file is required.");
		System.exit(1);
	}

	if (params.cramFile == null) {
		System.out.println("A CRAM input file is required. ");
		System.exit(1);
	}

	Log.setGlobalLogLevel(Log.LogLevel.INFO);

	ReferenceSequenceFile referenceSequenceFile = ReferenceSequenceFileFactory
			.getReferenceSequenceFile(params.reference);

	FileInputStream fis = new FileInputStream(params.cramFile);
	BufferedInputStream bis = new BufferedInputStream(fis);

	CRAMIterator iterator = new CRAMIterator(bis, new ReferenceSource(params.reference),
			ValidationStringency.STRICT);
	CramHeader cramHeader = iterator.getCramHeader();

	iterator.close();

	ProgressLogger progress = new ProgressLogger(log, 100000, "Validated Read");
	SamFileValidator v = new SamFileValidator(new PrintWriter(System.out), 1);
	final SamReader reader = SamReaderFactory.make().referenceSequence(params.reference).open(params.cramFile);
	List<SAMValidationError.Type> errors = new ArrayList<SAMValidationError.Type>();
	errors.add(SAMValidationError.Type.MATE_NOT_FOUND);
	// errors.add(Type.MISSING_TAG_NM);
	v.setErrorsToIgnore(errors);
	v.validateSamFileSummary(reader, ReferenceSequenceFileFactory.getReferenceSequenceFile(params.reference));
	log.info("Elapsed seconds: " + progress.getElapsedSeconds());
}