Search in sources :

Example 26 with ReferenceBases

use of org.broadinstitute.hellbender.utils.reference.ReferenceBases in project gatk by broadinstitute.

the class ReferenceTwoBitSourceUnitTest method testIntervalConversion.

@Test(dataProvider = "goodIntervals")
public void testIntervalConversion(ReferenceSource fastaRef, ReferenceSource twoBitRef, String intervalString) throws IOException {
    SimpleInterval interval = new SimpleInterval(intervalString);
    ReferenceBases expected = fastaRef.getReferenceBases(null, interval);
    ReferenceBases actual = twoBitRef.getReferenceBases(null, interval);
    Assert.assertEquals(actual, expected);
}
Also used : ReferenceBases(org.broadinstitute.hellbender.utils.reference.ReferenceBases) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) BaseTest(org.broadinstitute.hellbender.utils.test.BaseTest) Test(org.testng.annotations.Test)

Example 27 with ReferenceBases

use of org.broadinstitute.hellbender.utils.reference.ReferenceBases in project gatk by broadinstitute.

the class TandemRepeatUnitTest method testUsingVC.

@Test
public void testUsingVC() {
    // - [ref] / ATC from 20-20
    final String insLoc = "chr1";
    final int insLocStart = 2;
    final int insLocStop = 2;
    final byte[] refBytes = "GTATCATCATCGGA".getBytes();
    final Allele nullR = Allele.create("A", true);
    final Allele atc = Allele.create("AATC", false);
    // A*,ATC, context = ATC ATC ATC : (ATC)3 -> (ATC)4
    final VariantContext vc = new VariantContextBuilder("foo", insLoc, insLocStart, insLocStop, Arrays.asList(nullR, atc)).make();
    // we test that the interval from which the ReferenceContext is constructed does not need to exactly overlap
    // the VariantContext.  The annotation should be able to handle this.
    final SimpleInterval interval = new SimpleInterval(insLoc, insLocStart + 3, insLocStop + 4);
    final SimpleInterval interval1 = new SimpleInterval(insLoc, 1, refBytes.length);
    final ReferenceBases ref1 = new ReferenceBases(refBytes, interval1);
    final SAMSequenceDictionary dict = new SAMSequenceDictionary(Arrays.asList(new SAMSequenceRecord(insLoc, refBytes.length)));
    final ReferenceContext ref = new ReferenceContext(ReferenceDataSource.of(ref1, dict), interval, 20, 20);
    final InfoFieldAnnotation ann = new TandemRepeat();
    final Map<String, Object> a = ann.annotate(ref, vc, null);
    Assert.assertEquals(a.size(), 3);
    Assert.assertEquals(a.get(GATKVCFConstants.STR_PRESENT_KEY), true);
    Assert.assertEquals(a.get(GATKVCFConstants.REPEAT_UNIT_KEY), "ATC");
    //3 repeats in the reference, 4 in the variant
    Assert.assertEquals(a.get(GATKVCFConstants.REPEATS_PER_ALLELE_KEY), Arrays.asList(3, 4));
}
Also used : VariantContext(htsjdk.variant.variantcontext.VariantContext) SAMSequenceRecord(htsjdk.samtools.SAMSequenceRecord) SAMSequenceDictionary(htsjdk.samtools.SAMSequenceDictionary) ReferenceBases(org.broadinstitute.hellbender.utils.reference.ReferenceBases) Allele(htsjdk.variant.variantcontext.Allele) VariantContextBuilder(htsjdk.variant.variantcontext.VariantContextBuilder) ReferenceContext(org.broadinstitute.hellbender.engine.ReferenceContext) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) BaseTest(org.broadinstitute.hellbender.utils.test.BaseTest) Test(org.testng.annotations.Test)

Example 28 with ReferenceBases

use of org.broadinstitute.hellbender.utils.reference.ReferenceBases in project gatk by broadinstitute.

the class TandemRepeatUnitTest method testUsingVCNoRepeat.

@Test
public void testUsingVCNoRepeat() {
    // - [ref] / ATC from 20-20
    final String insLoc = "chr1";
    final int insLocStart = 6;
    final int insLocStop = 6;
    final byte[] refBytes = "GTATCATCATCGGA".getBytes();
    final Allele nullR = Allele.create("A", true);
    final Allele atc = Allele.create("AATC", false);
    // A*,ATC, context = ATC ATC ATC : (ATC)3 -> (ATC)4
    final VariantContext vc = new VariantContextBuilder("foo", insLoc, insLocStart, insLocStop, Arrays.asList(nullR, atc)).make();
    final SimpleInterval interval = new SimpleInterval(insLoc, insLocStart, insLocStop);
    final SimpleInterval interval1 = new SimpleInterval(insLoc, 1, refBytes.length);
    final ReferenceBases ref1 = new ReferenceBases(refBytes, interval1);
    final SAMSequenceDictionary dict = new SAMSequenceDictionary(Arrays.asList(new SAMSequenceRecord(insLoc, refBytes.length)));
    final ReferenceContext ref = new ReferenceContext(ReferenceDataSource.of(ref1, dict), interval, 0, 20);
    final InfoFieldAnnotation ann = new TandemRepeat();
    final Map<String, Object> a = ann.annotate(ref, vc, null);
    Assert.assertTrue(a.isEmpty());
}
Also used : VariantContext(htsjdk.variant.variantcontext.VariantContext) SAMSequenceRecord(htsjdk.samtools.SAMSequenceRecord) SAMSequenceDictionary(htsjdk.samtools.SAMSequenceDictionary) ReferenceBases(org.broadinstitute.hellbender.utils.reference.ReferenceBases) Allele(htsjdk.variant.variantcontext.Allele) VariantContextBuilder(htsjdk.variant.variantcontext.VariantContextBuilder) ReferenceContext(org.broadinstitute.hellbender.engine.ReferenceContext) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) BaseTest(org.broadinstitute.hellbender.utils.test.BaseTest) Test(org.testng.annotations.Test)

Example 29 with ReferenceBases

use of org.broadinstitute.hellbender.utils.reference.ReferenceBases in project gatk by broadinstitute.

the class TandemRepeatUnitTest method testUsingVCNotIndel.

@Test
public void testUsingVCNotIndel() {
    // - [ref] / ATC from 20-20
    String insLoc = "chr1";
    int insLocStart = 2;
    int insLocStop = 2;
    byte[] refBytes = "GTATCATCATCGGA".getBytes();
    Allele nullR = Allele.create("A", true);
    Allele atc = Allele.create("C", false);
    // A*,ATC, context = ATC ATC ATC : (ATC)3 -> (ATC)4
    VariantContext vc = new VariantContextBuilder("foo", insLoc, insLocStart, insLocStop, Arrays.asList(nullR, atc)).make();
    final SimpleInterval interval = new SimpleInterval("chr1", insLocStart, insLocStop);
    final String contigName = "chr1";
    final SimpleInterval interval1 = new SimpleInterval(contigName, 1, refBytes.length);
    final ReferenceBases ref1 = new ReferenceBases(refBytes, interval1);
    final SAMSequenceDictionary dict = new SAMSequenceDictionary(Arrays.asList(new SAMSequenceRecord(contigName, refBytes.length)));
    final ReferenceContext ref = new ReferenceContext(ReferenceDataSource.of(ref1, dict), interval, 0, 20);
    final InfoFieldAnnotation ann = new TandemRepeat();
    final Map<String, Object> a = ann.annotate(ref, vc, null);
    Assert.assertTrue(a.isEmpty());
}
Also used : VariantContext(htsjdk.variant.variantcontext.VariantContext) SAMSequenceRecord(htsjdk.samtools.SAMSequenceRecord) SAMSequenceDictionary(htsjdk.samtools.SAMSequenceDictionary) ReferenceBases(org.broadinstitute.hellbender.utils.reference.ReferenceBases) Allele(htsjdk.variant.variantcontext.Allele) VariantContextBuilder(htsjdk.variant.variantcontext.VariantContextBuilder) ReferenceContext(org.broadinstitute.hellbender.engine.ReferenceContext) SimpleInterval(org.broadinstitute.hellbender.utils.SimpleInterval) BaseTest(org.broadinstitute.hellbender.utils.test.BaseTest) Test(org.testng.annotations.Test)

Aggregations

ReferenceBases (org.broadinstitute.hellbender.utils.reference.ReferenceBases)29 SimpleInterval (org.broadinstitute.hellbender.utils.SimpleInterval)24 Test (org.testng.annotations.Test)15 BaseTest (org.broadinstitute.hellbender.utils.test.BaseTest)14 GATKRead (org.broadinstitute.hellbender.utils.read.GATKRead)10 SAMSequenceDictionary (htsjdk.samtools.SAMSequenceDictionary)6 ReferenceMultiSource (org.broadinstitute.hellbender.engine.datasources.ReferenceMultiSource)6 SAMSequenceRecord (htsjdk.samtools.SAMSequenceRecord)5 ReferenceContext (org.broadinstitute.hellbender.engine.ReferenceContext)5 ReferenceDataSource (org.broadinstitute.hellbender.engine.ReferenceDataSource)5 ReferenceMemorySource (org.broadinstitute.hellbender.engine.ReferenceMemorySource)5 JavaSparkContext (org.apache.spark.api.java.JavaSparkContext)4 ReadContextData (org.broadinstitute.hellbender.engine.ReadContextData)4 GATKVariant (org.broadinstitute.hellbender.utils.variant.GATKVariant)4 PipelineOptions (com.google.cloud.dataflow.sdk.options.PipelineOptions)3 Allele (htsjdk.variant.variantcontext.Allele)3 VariantContext (htsjdk.variant.variantcontext.VariantContext)3 VariantContextBuilder (htsjdk.variant.variantcontext.VariantContextBuilder)3 ArrayList (java.util.ArrayList)3 List (java.util.List)3