Examples with HopscotchSet org.broadinstitute.hellbender.tools.spark.utils.HopscotchSet used on opensource projects

Example 1 with HopscotchSet

use of org.broadinstitute.hellbender.tools.spark.utils.HopscotchSet in project gatk by broadinstitute.

the class PathSeqFilterSpark method doKmerFiltering.

@SuppressWarnings("unchecked")
private JavaRDD<GATKRead> doKmerFiltering(final JavaSparkContext ctx, final JavaRDD<GATKRead> reads) {
    final PipelineOptions options = getAuthenticatedGCSOptions();
    Input input = new Input(BucketUtils.openFile(KMER_LIB_PATH));
    Kryo kryo = new Kryo();
    kryo.setReferences(false);
    Set<SVKmer> kmerLibSet = (HopscotchSet<SVKmer>) kryo.readClassAndObject(input);
    return reads.filter(new ContainsKmerReadFilterSpark(ctx.broadcast(kmerLibSet), KMER_SIZE));
}

Example 2 with HopscotchSet

use of org.broadinstitute.hellbender.tools.spark.utils.HopscotchSet in project gatk by broadinstitute.

the class PathSeqKmerSpark method runTool.

/** Get the list of distinct kmers in the reference, and write them to a file as a HopScotchSet. */
@Override
protected void runTool(final JavaSparkContext ctx) {
    final SAMFileHeader hdr = getHeaderForReads();
    SAMSequenceDictionary dict = null;
    if (hdr != null)
        dict = hdr.getSequenceDictionary();
    final PipelineOptions options = getAuthenticatedGCSOptions();
    final ReferenceMultiSource referenceMultiSource = getReference();
    final List<SVKmer> kmerList = findKmers(ctx, KMER_SIZE, referenceMultiSource, options, dict);
    final HopscotchSet<SVKmer> kmerSet = new HopscotchSet<>(kmerList);
    final Output output = new Output(BucketUtils.createFile(OUTPUT_FILE));
    final Kryo kryo = new Kryo();
    kryo.setReferences(false);
    kryo.writeClassAndObject(output, kmerSet);
    output.close();
}

Example 3 with HopscotchSet

use of org.broadinstitute.hellbender.tools.spark.utils.HopscotchSet in project gatk by broadinstitute.

the class PathSeqKmerSparkIntegrationTest method test.

@SuppressWarnings("unchecked")
@Test(groups = "spark")
public void test() throws Exception {
    final File expectedFile = getTestFile("kmer.hss");
    final File ref = getTestFile("hg19mini.fasta");
    final File output = createTempFile("test", ".hss");
    if (!output.delete()) {
        Assert.fail();
    }
    final ArgumentsBuilder args = new ArgumentsBuilder();
    args.addFileArgument("reference", ref);
    args.addOutput(output);
    this.runCommandLine(args.getArgsArray());
    Input input_expected = new Input(FileUtils.openInputStream(expectedFile));
    Input input_test = new Input(FileUtils.openInputStream(output));
    Kryo kryo = new Kryo();
    kryo.setReferences(false);
    Set<SVKmer> expectedKmerLib = (HopscotchSet<SVKmer>) kryo.readClassAndObject(input_expected);
    Set<SVKmer> testKmerLib = (HopscotchSet<SVKmer>) kryo.readClassAndObject(input_test);
    Assert.assertEquals(expectedKmerLib, testKmerLib);
}