Java Code Examples for htsjdk.samtools.util.CollectionUtil#makeSet()

public RnaSeqMetricsCollector getCollector(final List<String> cellBarcodes) {
	List<SAMReadGroupRecord> readGroups =  getReadGroups(cellBarcodes);
	return new RnaSeqMtMetricsCollector(CollectionUtil.makeSet(MetricAccumulationLevel.READ_GROUP), readGroups,
               ribosomalBasesInitialValue, geneOverlapDetector, ribosomalSequenceOverlapDetector,
               ignoredSequenceIndices, MINIMUM_TRANSCRIPT_LENGTH, specificity, this.rnaFragPct, false,
               genesWithLongEnoughTranscripts);
}
 

/**
 * For each line in the MULTIPLEX_PARAMS file create a FastqRecordsWriter and put it in the sampleBarcodeFastqWriterMap map,
 * where the key to the map is the concatenation of all sampleBarcodes in order for the given line.
 */
private void populateWritersFromMultiplexParams() {
    final TabbedTextFileWithHeaderParser libraryParamsParser = new TabbedTextFileWithHeaderParser(MULTIPLEX_PARAMS);

    final Set<String> expectedColumnLabels = CollectionUtil.makeSet("OUTPUT_PREFIX");
    final List<String> sampleBarcodeColumnLabels = new ArrayList<>();
    for (int i = 1; i <= readStructure.sampleBarcodes.length(); i++) {
        sampleBarcodeColumnLabels.add("BARCODE_" + i);
    }

    expectedColumnLabels.addAll(sampleBarcodeColumnLabels);
    assertExpectedColumns(libraryParamsParser.columnLabels(), expectedColumnLabels);

    for (final TabbedTextFileWithHeaderParser.Row row : libraryParamsParser) {
        List<String> sampleBarcodeValues = null;

        if (!sampleBarcodeColumnLabels.isEmpty()) {
            sampleBarcodeValues = new ArrayList<>();
            for (final String sampleBarcodeLabel : sampleBarcodeColumnLabels) {
                sampleBarcodeValues.add(row.getField(sampleBarcodeLabel));
            }
        }

        final String key = (sampleBarcodeValues == null || sampleBarcodeValues.contains("N")) ? null : StringUtil.join("", sampleBarcodeValues);
        if (sampleBarcodeFastqWriterMap.containsKey(key)) {    //This will catch the case of having more than 1 line in a non-barcoded MULTIPLEX_PARAMS file
            throw new PicardException("Row for barcode " + key + " appears more than once in MULTIPLEX_PARAMS file " +
                    MULTIPLEX_PARAMS);
        }

        final FastqRecordsWriter writer = buildWriter(new File(row.getField("OUTPUT_PREFIX")));
        sampleBarcodeFastqWriterMap.put(key, writer);
    }
    if (sampleBarcodeFastqWriterMap.isEmpty()) {
        throw new PicardException("MULTIPLEX_PARAMS file " + MULTIPLEX_PARAMS + " does have any data rows.");
    }
    libraryParamsParser.close();
}
 

/**
 * Tests that sites with a het allele balance < 0.4 are marked as filtered out.
 */
@Test
public void testAbFiltering() throws Exception {
    final Set<String> fails = CollectionUtil.makeSet("tf2", "rs28566954", "rs28548431");
    final File out = testFiltering(INPUT, ".vcf.gz", 0.4, 0, 0, Double.MAX_VALUE);
    final ListMap<String, String> filters = slurpFilters(out);
    Assert.assertEquals(sorted(filters.keySet()), sorted(fails), "Failed sites did not match expected set of failed sites.");
}
 

/**
 * Tests that genotypes with DP < 18 are marked as failed, but not >= 18.
 */
@Test
public void testDpFiltering() throws Exception {
    final Set<String> fails = CollectionUtil.makeSet("rs71509448", "rs71628926", "rs13302979", "rs2710876");
    final File out = testFiltering(INPUT, ".vcf.gz", 0, 18, 0, Double.MAX_VALUE);
    final ListMap<String, String> filters = slurpFilters(out);
    Assert.assertEquals(sorted(filters.keySet()), sorted(fails), "Failed sites did not match expected set of failed sites.");
}
 

/**
 * Tests that genotypes with DP < 18 are marked as failed, but not >= 18.
 */
@Test
public void testDpFilteringToVcf() throws Exception {
    final Set<String> fails = CollectionUtil.makeSet("rs71509448", "rs71628926", "rs13302979", "rs2710876");
    final File out = testFiltering(INPUT, ".vcf", 0, 18, 0, Double.MAX_VALUE);
    final ListMap<String, String> filters = slurpFilters(out);
    Assert.assertEquals(sorted(filters.keySet()), sorted(fails), "Failed sites did not match expected set of failed sites.");
}
 

/**
 * Tests that genotypes with DP < 18 are marked as failed, but not >= 18.
 */
@Test(dataProvider = "goodInputVcfs")
public void testFsFiltering(final File input) throws Exception {
    final Set<String> fails = CollectionUtil.makeSet("rs13303033", "rs28548431", "rs2799066");
    final File out = testFiltering(input, ".vcf.gz", 0, 0, 0, 5.0d);
    final ListMap<String, String> filters = slurpFilters(out);
    Assert.assertEquals(sorted(filters.keySet()), sorted(fails), "Failed sites did not match expected set of failed sites.");
}
 

@Test
public void testCombinedFiltering() throws Exception {
    final TreeSet<String> fails = new TreeSet<String>(CollectionUtil.makeSet("rs13302979", "rs13303033", "rs2710876", "rs2799066", "rs28548431", "rs28566954", "rs71509448", "rs71628926", "tf2"));
    final File out = testFiltering(INPUT, ".vcf.gz", 0.4, 18, 22, 5.0d);
    final ListMap<String, String> filters = slurpFilters(out);
    Assert.assertEquals(new TreeSet<String>(filters.keySet()), fails, "Failed sites did not match expected set of failed sites.");
}
 

IntervalListInputType(final String... s) {
    applicableExtensions = CollectionUtil.makeSet(s);
}