Java SAMSequenceDictionary.assertSameDictionary方法代码示例

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
@Test
public void testBestSequenceDictionary_FromVariantReference() throws Exception {
    final GATKTool tool = new TestGATKToolWithFeatures();
    final CommandLineParser clp = new CommandLineArgumentParser(tool);
    final File vcfFile = new File(publicTestDir + "org/broadinstitute/hellbender/engine/example_variants_noSequenceDict.vcf");
    final String[] args = {"-V", vcfFile.getCanonicalPath(), "-R", hg19MiniReference};
    clp.parseArguments(System.out, args);
    tool.onStartup();
    // make sure we DON'T get the seq dictionary from the VCF index, and instead get the one from
    // the reference when its better
    final SAMSequenceDictionary toolDict = tool.getBestAvailableSequenceDictionary();
    Assert.assertFalse(toolDict.getSequences().stream().allMatch(seqRec -> seqRec.getSequenceLength() == 0));

    SAMSequenceDictionary refDict = new ReferenceFileSource(IOUtils.getPath(hg19MiniReference)).getSequenceDictionary();
    toolDict.assertSameDictionary(refDict);
    refDict.assertSameDictionary(toolDict);
    Assert.assertEquals(toolDict, refDict);
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
@Test
public void testBestSequenceDictionary_fromReads() throws Exception {
    final GATKTool tool = new TestGATKToolWithReads();
    final CommandLineParser clp = new CommandLineArgumentParser(tool);
    final File bamFile = new File(publicTestDir + "org/broadinstitute/hellbender/engine/reads_data_source_test1.bam");
    final String[] args = {"-I", bamFile.getCanonicalPath()};
    clp.parseArguments(System.out, args);
    tool.onStartup();
    //read the dict back in and compare to bam dict
    final SAMSequenceDictionary toolDict = tool.getBestAvailableSequenceDictionary();
    try(final SamReader open = SamReaderFactory.makeDefault().open(bamFile)) {
        final SAMSequenceDictionary bamDict = open.getFileHeader().getSequenceDictionary();
        toolDict.assertSameDictionary(bamDict);
        bamDict.assertSameDictionary(toolDict);
        Assert.assertEquals(toolDict, bamDict);
    }
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
@Test
public void testBestSequenceDictionary_fromReadsAndReference() throws Exception {
    final GATKTool tool = new TestGATKToolWithReads();
    final CommandLineParser clp = new CommandLineArgumentParser(tool);
    final File bamFile = new File(publicTestDir + "org/broadinstitute/hellbender/engine/reads_data_source_test1.bam");
    final String fastaFile =   hg19MiniReference;
    final String[] args = {"-I", bamFile.getCanonicalPath(),
                           "-R", fastaFile};
    clp.parseArguments(System.out, args);
    tool.onStartup();
    //read the dict back in and compare to reference dict
    final SAMSequenceDictionary toolDict = tool.getBestAvailableSequenceDictionary();
    final SAMSequenceDictionary fastaDict = new IndexedFastaSequenceFile(new File(fastaFile)).getSequenceDictionary();
    toolDict.assertSameDictionary(fastaDict);
    fastaDict.assertSameDictionary(toolDict);

    Assert.assertEquals(toolDict, fastaDict);
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
@Test
public void testBestSequenceDictionary_fromVariants() throws Exception {
    final GATKTool tool = new TestGATKToolWithFeatures();
    final CommandLineParser clp = new CommandLineArgumentParser(tool);
    final File vcfFile = new File(publicTestDir + "org/broadinstitute/hellbender/engine/feature_data_source_test_withSequenceDict.vcf");
    final String[] args = {"--mask", vcfFile.getCanonicalPath()};
    clp.parseArguments(System.out, args);
    tool.onStartup();
    //read the dict back in and compare to vcf dict
    final SAMSequenceDictionary toolDict = tool.getBestAvailableSequenceDictionary();
    try(final VCFFileReader reader = new VCFFileReader(vcfFile)) {
        final SAMSequenceDictionary vcfDict = reader.getFileHeader().getSequenceDictionary();
        toolDict.assertSameDictionary(vcfDict);
        vcfDict.assertSameDictionary(toolDict);
        Assert.assertEquals(toolDict, vcfDict);
    }
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
private void collectFileReadersAndHeaders(final List<String> sampleList, SAMSequenceDictionary samSequenceDictionary) {
    for (final File input : INPUT) {
        final VCFFileReader in = new VCFFileReader(input, false);
        final VCFHeader header = in.getFileHeader();
        final SAMSequenceDictionary dict = in.getFileHeader().getSequenceDictionary();
        if (dict == null || dict.isEmpty()) {
            if (null == samSequenceDictionary) {
                throw new IllegalArgumentException("Sequence dictionary was missing or empty for the VCF: " + input.getAbsolutePath() + " Please add a sequence dictionary to this VCF or specify SEQUENCE_DICTIONARY.");
            }
            header.setSequenceDictionary(samSequenceDictionary);
        } else {
            if (null == samSequenceDictionary) {
                samSequenceDictionary = dict;
            } else {
                try {
                    samSequenceDictionary.assertSameDictionary(dict);
                } catch (final AssertionError e) {
                    throw new IllegalArgumentException(e);
                }
            }
        }
        if (sampleList.isEmpty()) {
            sampleList.addAll(header.getSampleNamesInOrder());
        } else {
            if (!sampleList.equals(header.getSampleNamesInOrder())) {
                throw new IllegalArgumentException("Input file " + input.getAbsolutePath() + " has sample names that don't match the other files.");
            }
        }
        inputReaders.add(in);
        inputHeaders.add(header);
    }
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
private void assertFastaDictionaryContent(final Path dictPath, final SAMSequenceDictionary dictionary)
        throws IOException, GeneralSecurityException, URISyntaxException {
    final ReadsDataSource readsDataSource = new ReadsDataSource(dictPath);
    final SAMFileHeader actualHeader = readsDataSource.getHeader();
    final SAMSequenceDictionary actualDictionary = actualHeader.getSequenceDictionary();
    dictionary.assertSameDictionary(actualDictionary);
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
/** Validates that all headers contain the same set of genotyped samples and that files are in order by position of first record. */
private static void assertSameSamplesAndValidOrdering(final List<File> inputFiles) {
    final VCFHeader header = new VCFFileReader(inputFiles.get(0), false).getFileHeader();
    final SAMSequenceDictionary dict = header.getSequenceDictionary();
    final VariantContextComparator comparator = new VariantContextComparator(header.getSequenceDictionary());
    final List<String> samples = header.getGenotypeSamples();

    File lastFile = null;
    VariantContext lastContext = null;

    for (final File f : inputFiles) {
        final VCFFileReader in = new VCFFileReader(f, false);
        try {
            dict.assertSameDictionary(in.getFileHeader().getSequenceDictionary());
        } catch (final AssertionError e) {
            log.error("File #1: " + inputFiles.get(0));
            log.error("File #2: " + f);
            throw e;
        }
        final List<String> theseSamples = in.getFileHeader().getGenotypeSamples();

        if (!samples.equals(theseSamples)) {
            final SortedSet<String> s1 = new TreeSet<String>(samples);
            final SortedSet<String> s2 = new TreeSet<String>(theseSamples);
            s1.removeAll(theseSamples);
            s2.removeAll(samples);

            throw new IllegalArgumentException("VCFs do not have identical sample lists." +
                    " Samples unique to first file: " + s1 + ". Samples unique to " + f.getAbsolutePath() + ": " + s2 + ".");
        }

        final CloseableIterator<VariantContext> variantIterator = in.iterator();
        if (variantIterator.hasNext()) {
            final VariantContext currentContext = variantIterator.next();
            if (lastContext != null && comparator.compare(lastContext, currentContext) >= 0) {
                throw new IllegalArgumentException("First record in file " + f.getAbsolutePath() + " is not after first record in " +
                        "previous file " + lastFile.getAbsolutePath());
            }

            lastContext = currentContext;
            lastFile    = f;
        }

        CloserUtil.close(in);
    }
}
 

import htsjdk.samtools.SAMSequenceDictionary; //导入方法依赖的package包/类
/** Validates that all headers contain the same set of genotyped samples and that files are in order by position of first record. */
private static void assertSameSamplesAndValidOrdering(final List<Path> inputFiles, final boolean disableContigOrderingCheck) {
    final VCFHeader firstHeader = getHeader(inputFiles.get(0));
    final SAMSequenceDictionary dict = firstHeader.getSequenceDictionary();
    if ( dict == null) {
        throw new UserException.BadInput("The first VCF specified is missing the required sequence dictionary. " +
                                                 "This is required to perform validation.  You can skip this validation " +
                                                 "using --"+IGNORE_SAFETY_CHECKS_LONG_NAME +" but ignoring safety checks " +
                                                 "can result in invalid output.");
    }
    final VariantContextComparator comparator = new VariantContextComparator(dict);
    final List<String> samples = firstHeader.getGenotypeSamples();

    Path lastFile = null;
    VariantContext lastContext = null;

    for (final Path f : inputFiles) {
        final FeatureReader<VariantContext> in = getReaderFromVCFUri(f, 0);
        final VCFHeader header = (VCFHeader)in.getHeader();
        dict.assertSameDictionary(header.getSequenceDictionary());
        final List<String> theseSamples = header.getGenotypeSamples();

        if (!samples.equals(theseSamples)) {
            final SortedSet<String> s1 = new TreeSet<>(samples);
            final SortedSet<String> s2 = new TreeSet<>(theseSamples);
            s1.removeAll(theseSamples);
            s2.removeAll(samples);

            throw new IllegalArgumentException("VCFs do not have identical sample lists." +
                    " Samples unique to first file: " + s1 + ". Samples unique to " + f.toUri().toString() + ": " + s2 + ".");
        }

        try(final CloseableIterator<VariantContext> variantIterator = in.iterator()) {
            if (variantIterator.hasNext()) {
                final VariantContext currentContext = variantIterator.next();
                if (lastContext != null) {
                    if ( disableContigOrderingCheck ) {
                        if ( lastContext.getContig().equals(currentContext.getContig()) && lastContext.getStart() >= currentContext.getStart() ) {
                            throw new IllegalArgumentException(
                                    "First record in file " + f.toUri().toString() + " is not after first record in " +
                                            "previous file " + lastFile.toUri().toString());
                        }
                    }
                    else {
                        if ( comparator.compare(lastContext, currentContext) >= 0 ) {
                            throw new IllegalArgumentException(
                                    "First record in file " + f.toUri().toString() + " is not after first record in " +
                                            "previous file " + lastFile.toUri().toString());
                        }
                    }
                }

                lastContext = currentContext;
                lastFile = f;
            }
        } catch (final IOException e) {
            throw new UserException.CouldNotReadInputFile(f, e.getMessage(), e);
        }

        CloserUtil.close(in);
    }
}