Examples with SamReader htsjdk.samtools.SamReader used on opensource projects

Example 51 with SamReader

use of htsjdk.samtools.SamReader in project gridss by PapenfussLab.

the class SplitReadRealigner method createSupplementaryAlignments.

public void createSupplementaryAlignments(StreamingAligner aligner, File input, File output) throws IOException {
    SplitReadFastqExtractor rootExtractor = new SplitReadFastqExtractor(false, minSoftClipLength, minSoftClipQuality, isProcessSecondaryAlignments(), eidgen);
    SplitReadFastqExtractor recursiveExtractor = new SplitReadFastqExtractor(true, minSoftClipLength, minSoftClipQuality, false, eidgen);
    Map<String, SplitReadRealignmentInfo> realignments = new HashMap<>();
    try (SamReader reader = readerFactory.open(input)) {
        SAMFileHeader header = reader.getFileHeader().clone();
        header.setSortOrder(SortOrder.unsorted);
        try (SAMFileWriter writer = writerFactory.makeSAMOrBAMWriter(header, true, output)) {
            try (AsyncBufferedIterator<SAMRecord> bufferedIt = new AsyncBufferedIterator<>(reader.iterator(), input.getName())) {
                while (bufferedIt.hasNext()) {
                    SAMRecord r = bufferedIt.next();
                    processInputRecord(aligner, rootExtractor, realignments, writer, r);
                    while (aligner.hasAlignmentRecord()) {
                        processAlignmentRecord(aligner, recursiveExtractor, realignments, writer);
                    }
                }
                // flush out all realignments
                aligner.flush();
                while (aligner.hasAlignmentRecord()) {
                    // perform nested realignment
                    while (aligner.hasAlignmentRecord()) {
                        processAlignmentRecord(aligner, recursiveExtractor, realignments, writer);
                    }
                    aligner.flush();
                }
            }
        }
    }
    assert (realignments.size() == 0);
}

Example 52 with SamReader

use of htsjdk.samtools.SamReader in project gridss by PapenfussLab.

the class ComputeSamTags method doWork.

@Override
protected int doWork() {
    log.debug("Setting language-neutral locale");
    java.util.Locale.setDefault(Locale.ROOT);
    validateParameters();
    SamReaderFactory readerFactory = SamReaderFactory.make();
    SAMFileWriterFactory writerFactory = new SAMFileWriterFactory();
    try {
        try (SamReader reader = readerFactory.open(INPUT)) {
            SAMFileHeader header = reader.getFileHeader();
            if (!ASSUME_SORTED) {
                if (header.getSortOrder() != SortOrder.queryname) {
                    log.error("INPUT is not sorted by queryname. " + "ComputeSamTags requires that reads with the same name be sorted together. " + "If the input file satisfies this constraint (the output from many aligners do)," + " this check can be disabled with the ASSUME_SORTED option.");
                    return -1;
                }
            }
            try (SAMRecordIterator it = reader.iterator()) {
                File tmpoutput = gridss.Defaults.OUTPUT_TO_TEMP_FILE ? FileSystemContext.getWorkingFileFor(OUTPUT, "gridss.tmp.ComputeSamTags.") : OUTPUT;
                try (SAMFileWriter writer = writerFactory.makeSAMOrBAMWriter(header, true, tmpoutput)) {
                    compute(it, writer, getReference(), TAGS, SOFTEN_HARD_CLIPS, FIX_MATE_INFORMATION, RECALCULATE_SA_SUPPLEMENTARY, INPUT.getName() + "-");
                }
                if (tmpoutput != OUTPUT) {
                    FileHelper.move(tmpoutput, OUTPUT, true);
                }
            }
        }
    } catch (IOException e) {
        log.error(e);
        return -1;
    }
    return 0;
}

Example 53 with SamReader

use of htsjdk.samtools.SamReader in project gridss by PapenfussLab.

the class ByReadNameSinglePassSamProgram method makeItSo.

public static void makeItSo(final File input, final File referenceSequence, final boolean assumeSorted, final long stopAfter, final Collection<ByReadNameSinglePassSamProgram> programs) throws FileNotFoundException {
    // Setup the standard inputs
    IOUtil.assertFileIsReadable(input);
    SamReader in = SamReaderFactory.makeDefault().referenceSequence(referenceSequence).open(input);
    // Optionally load up the reference sequence and double check sequence dictionaries
    final ReferenceLookup lookup;
    if (referenceSequence == null) {
        lookup = null;
    } else {
        IOUtil.assertFileIsReadable(referenceSequence);
        lookup = new TwoBitBufferedReferenceSequenceFile(new IndexedFastaSequenceFile(referenceSequence));
        if (!in.getFileHeader().getSequenceDictionary().isEmpty()) {
            SequenceUtil.assertSequenceDictionariesEqual(in.getFileHeader().getSequenceDictionary(), lookup.getSequenceDictionary());
        }
    }
    // Check on the sort order of the BAM file
    final SortOrder sort = in.getFileHeader().getSortOrder();
    if (sort != SortOrder.queryname) {
        if (assumeSorted) {
            log.warn("File reports sort order '" + sort + "', assuming it's queryname sorted anyway.");
        } else {
            throw new PicardException("File " + input.getAbsolutePath() + " should be queryname sorted but " + "the header says the sort order is " + sort + ". If you believe the file " + "to be queryname sorted you may pass ASSUME_SORTED=true");
        }
    }
    for (final ByReadNameSinglePassSamProgram program : programs) {
        program.setReference(lookup);
        program.setup(in.getFileHeader(), input);
    }
    final ProgressLogger progress = new ProgressLogger(log);
    final SAMRecordIterator rawit = in.iterator();
    final CloseableIterator<SAMRecord> it = new AsyncBufferedIterator<SAMRecord>(rawit, "ByReadNameSinglePassSamProgram " + input.getName());
    try {
        List<SAMRecord> currentRecords = new ArrayList<>();
        String currentReadName = null;
        while (it.hasNext()) {
            SAMRecord r = it.next();
            String readname = r.getReadName();
            // if read name we have to just treat it as a single read
            if (readname == null || !readname.equals(currentReadName)) {
                if (currentRecords.size() > 0) {
                    for (final ByReadNameSinglePassSamProgram program : programs) {
                        program.acceptFragment(currentRecords, lookup);
                    }
                }
                currentRecords.clear();
                currentReadName = readname;
                if (stopAfter > 0 && progress.getCount() >= stopAfter) {
                    break;
                }
            }
            currentRecords.add(r);
            progress.record(r);
        }
        if (currentRecords.size() > 0) {
            for (final ByReadNameSinglePassSamProgram program : programs) {
                program.acceptFragment(currentRecords, lookup);
            }
        }
    } finally {
        CloserUtil.close(it);
        CloserUtil.close(rawit);
        CloserUtil.close(in);
    }
    for (final ByReadNameSinglePassSamProgram program : programs) {
        program.finish();
    }
}

Example 54 with SamReader

use of htsjdk.samtools.SamReader in project gridss by PapenfussLab.

the class MultipleSamFileCommandLineProgram method getCategoryLabelFor.

private String getCategoryLabelFor(File file) {
    String label = file.getName();
    if (file.exists() && getContext().getConfig().useReadGroupSampleNameCategoryLabel) {
        try (SamReader reader = SamReaderFactory.makeDefault().open(file)) {
            SAMFileHeader header = reader.getFileHeader();
            if (header.getReadGroups().size() == 1) {
                String sampleName = header.getReadGroups().get(0).getSample();
                if (sampleName != null && sampleName.length() > 0) {
                    label = sampleName;
                    log.info(String.format("Using Read Group sample name %s for file %s", label, file));
                }
            }
        } catch (IOException e) {
            log.debug(e, "Unable to load read group headers for ", file);
        }
    }
    return label;
}

Example 55 with SamReader

use of htsjdk.samtools.SamReader in project gridss by PapenfussLab.

the class MultipleSamFileCommandLineProgram method ensureDictionariesMatch.

public void ensureDictionariesMatch() throws IOException {
    ReferenceSequenceFile ref = null;
    try {
        ref = ReferenceSequenceFileFactory.getReferenceSequenceFile(REFERENCE_SEQUENCE);
        SAMSequenceDictionary dictionary = ref.getSequenceDictionary();
        final SamReaderFactory samFactory = SamReaderFactory.makeDefault();
        for (File f : INPUT) {
            SamReader reader = null;
            try {
                reader = samFactory.open(f);
                SequenceUtil.assertSequenceDictionariesEqual(reader.getFileHeader().getSequenceDictionary(), dictionary, f, REFERENCE_SEQUENCE);
            } catch (htsjdk.samtools.util.SequenceUtil.SequenceListsDifferException e) {
                log.error("Reference genome used by ", f, " does not match reference genome ", REFERENCE_SEQUENCE, ". ", "The reference supplied must match the reference used for every input.");
                throw e;
            } finally {
                if (reader != null)
                    reader.close();
            }
        }
    } finally {
        if (ref != null)
            ref.close();
    }
}