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

Example 1 with QueryInterval

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

the class IntervalUtilsUnitTest method testConvertSimpleIntervalToQueryInterval.

@Test
public void testConvertSimpleIntervalToQueryInterval() {
    final SAMSequenceRecord contigRecord = new SAMSequenceRecord("1", 100);
    final SAMSequenceDictionary dictionary = new SAMSequenceDictionary(Arrays.asList(contigRecord));
    final SimpleInterval originalInterval = new SimpleInterval("1", 5, 10);
    final QueryInterval convertedInterval = IntervalUtils.convertSimpleIntervalToQueryInterval(originalInterval, dictionary);
    Assert.assertEquals(convertedInterval.referenceIndex, 0);
    Assert.assertEquals(convertedInterval.start, 5);
    Assert.assertEquals(convertedInterval.end, 10);
}

Example 2 with QueryInterval

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

the class NioBam method getReads.

/** Parses the BAM file into SAMRecords. Will be distributed onto at least 'numPartitions' partitions. **/
public JavaRDD<SAMRecord> getReads(JavaSparkContext ctx, int numPartitions) {
    try {
        Path bamPath = IOUtils.getPath(bam);
        ChannelAsSeekableStream bamOverNIO = new ChannelAsSeekableStream(Files.newByteChannel(bamPath), bamPath.toString());
        final byte[] index = getIndex();
        SeekableStream indexInMemory = new ByteArraySeekableStream(index);
        SamReader bam3 = SamReaderFactory.makeDefault().validationStringency(ValidationStringency.LENIENT).enable(SamReaderFactory.Option.CACHE_FILE_BASED_INDEXES).open(SamInputResource.of(bamOverNIO).index(indexInMemory));
        List<QueryInterval> chunks = getAllChunksBalanced(bam3, numPartitions);
        // Ideally we'd get exactly the number of chunks the user is asking for, but until then...
        logger.debug("We got: " + chunks.size() + " chunks.");
        return ctx.parallelize(chunks, chunks.size()).flatMap(qi -> new ReadsIterable(bam, index, qi).iterator());
    } catch (IOException e) {
        throw new GATKException("I/O error loading reads", e);
    }
}

Example 3 with QueryInterval

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

the class NioBam method getChunksBalanced.

private static List<QueryInterval> getChunksBalanced(SamReader bam, int sequenceIndex, int retCount) {
    List<QueryInterval> ret = new ArrayList<>();
    BAMIndex index = bam.indexing().getIndex();
    SAMFileHeader header = bam.getFileHeader();
    SAMSequenceRecord s = header.getSequence(sequenceIndex);
    long totalLength = chunksLength(getChunks(index, sequenceIndex, 1, s.getSequenceLength() + 1));
    if (totalLength == 0) {
        return ret;
    }
    int sofar = 0;
    long targetLength = totalLength / retCount;
    int end = s.getSequenceLength();
    int step = s.getSequenceLength() / (100 * retCount);
    if (step < 1)
        step = 1;
    int start = 1;
    for (int j = step; j < end; j += step) {
        if (j > end)
            j = end;
        List<Chunk> candidate = getChunks(index, sequenceIndex, start, j);
        long size = chunksLength(candidate);
        if (size < targetLength) {
            // not big enough yet
            continue;
        }
        if (size > targetLength * 2) {
        // too large, search for a good separation point
        // TODO
        }
        // good, emit.
        ret.add(new QueryInterval(sequenceIndex, start, j + 1));
        start = j;
        sofar += size;
        if (ret.size() < retCount) {
            targetLength = (totalLength - sofar) / (retCount - ret.size());
        } else {
            targetLength = totalLength / retCount;
        }
    }
    return ret;
}

Example 4 with QueryInterval

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

the class NioBam method getAllChunksBalanced.

// this isn't very good yet, ideally we want just this number of query intervals, not per-contig.
private static List<QueryInterval> getAllChunksBalanced(SamReader bam, int countPerContig) {
    List<QueryInterval> ret = new ArrayList<>();
    SAMFileHeader header = bam.getFileHeader();
    for (SAMSequenceRecord s : header.getSequenceDictionary().getSequences()) {
        ret.addAll(getChunksBalanced(bam, s.getSequenceIndex(), countPerContig));
    }
    return ret;
}