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

Example 66 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project gatk by broadinstitute.

the class SampleCollectionUnitTest method testReadGroupIds.

@Test(dataProvider = "samFileHeaderData", dependsOnMethods = "testCreation")
public void testReadGroupIds(final SAMFileHeader header) {
    final SampleCollection sampleCollection = new SampleCollection(header);
    final List<String> readGroupIds = sampleCollection.readGroups();
    Assert.assertEquals(readGroupIds.size(), header.getReadGroups().size());
    final List<String> expectedReadGroupIds = new ArrayList<>();
    header.getReadGroups().forEach(rg -> expectedReadGroupIds.add(rg.getId()));
    Collections.sort(expectedReadGroupIds, (a, b) -> {
        final SAMReadGroupRecord arg = header.getReadGroup(a);
        final SAMReadGroupRecord brg = header.getReadGroup(b);
        final String aSampleId = arg.getSample();
        final String bSampleId = brg.getSample();
        if (Objects.equals(aSampleId, bSampleId)) {
            return arg.getId().compareTo(brg.getId());
        } else if (aSampleId == null) {
            return 1;
        } else if (bSampleId == null) {
            return -1;
        } else {
            return aSampleId.compareTo(bSampleId);
        }
    });
    Assert.assertEquals(readGroupIds, expectedReadGroupIds);
}

Example 67 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project gatk by broadinstitute.

the class LocusIteratorByStateUnitTest method testLIBS_NotHoldingTooManyReads.

@Test(enabled = true, dataProvider = "LIBS_NotHoldingTooManyReads")
public void testLIBS_NotHoldingTooManyReads(final int nReadsPerLocus, final int downsampleTo, final int payloadInBytes) {
    logger.warn(String.format("testLIBS_NotHoldingTooManyReads %d %d %d", nReadsPerLocus, downsampleTo, payloadInBytes));
    final int readLength = 10;
    final SAMFileHeader header = ArtificialReadUtils.createArtificialSamHeader(1, 1, 100000);
    final int nSamples = 1;
    final List<String> samples = new ArrayList<>(nSamples);
    for (int i = 0; i < nSamples; i++) {
        final SAMReadGroupRecord rg = new SAMReadGroupRecord("rg" + i);
        final String sample = "sample" + i;
        samples.add(sample);
        rg.setSample(sample);
        rg.setPlatform(NGSPlatform.ILLUMINA.getDefaultPlatform());
        header.addReadGroup(rg);
    }
    final boolean downsample = downsampleTo != -1;
    final DownsamplingMethod downsampler = downsample ? new DownsamplingMethod(DownsampleType.BY_SAMPLE, downsampleTo, null) : new DownsamplingMethod(DownsampleType.NONE, null, null);
    final WeakReadTrackingIterator iterator = new WeakReadTrackingIterator(nReadsPerLocus, readLength, payloadInBytes, header);
    final LocusIteratorByState li;
    li = new LocusIteratorByState(iterator, downsampler, false, samples, header, true);
    while (li.hasNext()) {
        final AlignmentContext next = li.next();
        Assert.assertTrue(next.getBasePileup().size() <= downsampleTo, "Too many elements in pileup " + next);
    // TODO -- assert that there are <= X reads in memory after GC for some X
    }
}

Example 68 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project polyGembler by c-zhou.

the class SAMtools method run.

@Override
public void run() {
    // TODO Auto-generated method stub
    final SamReaderFactory factory = SamReaderFactory.makeDefault().enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS, SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS).validationStringency(ValidationStringency.SILENT);
    final SamReader inputSam = factory.open(new File(mySamFile));
    samHeader = inputSam.getFileHeader();
    samHeader.setSortOrder(SortOrder.unsorted);
    SAMRecordIterator iter = inputSam.iterator();
    Set<Entry<String, String>> attr = samHeader.getAttributes();
    List<SAMReadGroupRecord> rgs = samHeader.getReadGroups();
    SAMReadGroupRecord rg = new SAMReadGroupRecord("cz1");
    rg.setSample("cz1");
    samHeader.addReadGroup(rg);
    // samHeader.setAttribute("RG", "cz1");
    final SAMFileWriter outSam = new SAMFileWriterFactory().makeSAMOrBAMWriter(samHeader, true, new File(myOutput));
    for (int i = 0; i < 100; i++) {
        SAMRecord record = iter.next();
        List<SAMTagAndValue> tags = record.getAttributes();
        record.setAttribute("RG", "cz1");
        List<SAMTagAndValue> tags2 = record.getAttributes();
        outSam.addAlignment(record);
    }
    myLogger.info("exit...");
    try {
        inputSam.close();
    } catch (IOException e) {
        // TODO Auto-generated catch block
        e.printStackTrace();
    }
    outSam.close();
}

Example 69 with SAMReadGroupRecord

use of htsjdk.samtools.SAMReadGroupRecord in project polyGembler by c-zhou.

the class SamFileSplit method run.

@Override
public void run() {
    // TODO Auto-generated method stub
    Utils.makeOutputDir(bam_out);
    final File[] beds = new File(bed_in).listFiles();
    final String[] out_prefix = new String[beds.length];
    for (int i = 0; i < beds.length; i++) {
        out_prefix[i] = bam_out + "/" + beds[i].getName().replaceAll(".bed$", "");
        Utils.makeOutputDir(out_prefix[i]);
    }
    final File[] bams = new File(bam_in).listFiles(new FilenameFilter() {

        @Override
        public boolean accept(File dir, String name) {
            return name.endsWith(".bam");
        }
    });
    this.initial_thread_pool();
    for (File bam : bams) {
        executor.submit(new Runnable() {

            private File bam;

            @Override
            public void run() {
                // TODO Auto-generated method stub
                try {
                    final SamReaderFactory factory = SamReaderFactory.makeDefault().enable(SamReaderFactory.Option.INCLUDE_SOURCE_IN_RECORDS, SamReaderFactory.Option.VALIDATE_CRC_CHECKSUMS).validationStringency(ValidationStringency.SILENT);
                    final SamReader inputSam = factory.open(bam);
                    final SAMFileHeader header = inputSam.getFileHeader();
                    final SAMRecordIterator iter = inputSam.iterator();
                    final SAMSequenceDictionary seqdic = header.getSequenceDictionary();
                    final SAMFileWriter[] outputSam = new SAMFileWriter[beds.length];
                    final SAMSequenceDictionary[] seqdics = new SAMSequenceDictionary[beds.length];
                    final Map<String, Integer> outMap = new HashMap<String, Integer>();
                    final String out = bam.getName();
                    for (int i = 0; i < beds.length; i++) {
                        Set<String> bed_seq = new HashSet<String>();
                        String tmp;
                        BufferedReader br = new BufferedReader(new FileReader(beds[i]));
                        String line;
                        while ((line = br.readLine()) != null) {
                            tmp = line.split("\\s+")[0];
                            bed_seq.add(tmp);
                            outMap.put(tmp, i);
                        }
                        br.close();
                        final SAMFileHeader header_i = new SAMFileHeader();
                        final SAMSequenceDictionary seqdic_i = new SAMSequenceDictionary();
                        header_i.setAttribute("VN", header.getAttribute("VN"));
                        header_i.setAttribute("SO", header.getAttribute("SO"));
                        List<SAMSequenceRecord> seqs = seqdic.getSequences();
                        for (SAMSequenceRecord seq : seqs) if (bed_seq.contains(seq.getSequenceName()))
                            seqdic_i.addSequence(seq);
                        header_i.setSequenceDictionary(seqdic_i);
                        for (SAMReadGroupRecord rg : header.getReadGroups()) header_i.addReadGroup(rg);
                        for (SAMProgramRecord pg : header.getProgramRecords()) header_i.addProgramRecord(pg);
                        outputSam[i] = new SAMFileWriterFactory().makeSAMOrBAMWriter(header_i, true, new File(out_prefix[i] + "/" + out));
                        seqdics[i] = seqdic_i;
                    }
                    Set<String> refs = outMap.keySet();
                    String ref;
                    int f;
                    while (iter.hasNext()) {
                        SAMRecord rec = iter.next();
                        if (refs.contains(ref = rec.getReferenceName())) {
                            f = outMap.get(ref);
                            rec.setReferenceIndex(seqdics[f].getSequenceIndex(ref));
                            outputSam[f].addAlignment(rec);
                        }
                    }
                    iter.close();
                    inputSam.close();
                    for (int i = 0; i < outputSam.length; i++) outputSam[i].close();
                    myLogger.info(out + " return true");
                } catch (Exception e) {
                    Thread t = Thread.currentThread();
                    t.getUncaughtExceptionHandler().uncaughtException(t, e);
                    e.printStackTrace();
                    executor.shutdown();
                    System.exit(1);
                }
            }

            public Runnable init(File bam) {
                this.bam = bam;
                return (this);
            }
        }.init(bam));
    }
    this.waitFor();
}

Example 70 with SAMReadGroupRecord

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

the class TestHelper method getHeader.

public static SAMFileHeader getHeader() {
    SAMFileHeader header = new SAMFileHeader();
    header.setSequenceDictionary(getSequenceDictionary());
    SAMReadGroupRecord rg = new SAMReadGroupRecord("RG");
    rg.setSample("sample");
    header.addReadGroup(rg);
    return header;
}