zzh
/
gatk-3.8
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Adding ability to exclude annotations from the VA and UG lists. As described in the docs, this argument trumps all others (including -all) so that we can get around the SnpEff issue brought up by Menachem. Added integration test for it.

This commit is contained in:
Eric Banks 2011-10-12 15:44:54 -04:00
parent e53a952aeb
commit 9aecd50473
4 changed files with 62 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

14
public/java/src/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotator.java
View File

@ -132,6 +132,13 @@ public class VariantAnnotator extends RodWalker<Integer, Integer> implements Ann
@Argument(fullName="annotation", shortName="A", doc="One or more specific annotations to apply to variant calls", required=false)
protected List<String> annotationsToUse = new ArrayList<String>();
/**
* Note that this argument has higher priority than the -A or -G arguments,
* so annotations will be excluded even if they are explicitly included with the other options.
*/
@Argument(fullName="excludeAnnotation", shortName="XA", doc="One or more specific annotations to exclude", required=false)
protected List<String> annotationsToExclude = new ArrayList<String>();
/**
* See the -list argument to view available groups.
*/
@ -148,6 +155,9 @@ public class VariantAnnotator extends RodWalker<Integer, Integer> implements Ann
@Argument(fullName="expression", shortName="E", doc="One or more specific expressions to apply to variant calls; see documentation for more details", required=false)
protected List<String> expressionsToUse = new ArrayList<String>();
/**
* Note that the -XL argument can be used along with this one to exclude annotations.
*/
@Argument(fullName="useAllAnnotations", shortName="all", doc="Use all possible annotations (not for the faint of heart)", required=false)
protected Boolean USE_ALL_ANNOTATIONS = false;
@ -209,9 +219,9 @@ public class VariantAnnotator extends RodWalker<Integer, Integer> implements Ann
}
if ( USE_ALL_ANNOTATIONS )
engine = new VariantAnnotatorEngine(this, getToolkit());
engine = new VariantAnnotatorEngine(annotationsToExclude, this, getToolkit());
else
engine = new VariantAnnotatorEngine(annotationGroupsToUse, annotationsToUse, this, getToolkit());
engine = new VariantAnnotatorEngine(annotationGroupsToUse, annotationsToUse, annotationsToExclude, this, getToolkit());
engine.initializeExpressions(expressionsToUse);
// setup the header fields

29
public/java/src/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotatorEngine.java
View File

@ -73,19 +73,20 @@ public class VariantAnnotatorEngine {
}
// use this constructor if you want all possible annotations
public VariantAnnotatorEngine(AnnotatorCompatibleWalker walker, GenomeAnalysisEngine toolkit) {
public VariantAnnotatorEngine(List<String> annotationsToExclude, AnnotatorCompatibleWalker walker, GenomeAnalysisEngine toolkit) {
this.walker = walker;
this.toolkit = toolkit;
requestedInfoAnnotations = AnnotationInterfaceManager.createAllInfoFieldAnnotations();
requestedGenotypeAnnotations = AnnotationInterfaceManager.createAllGenotypeAnnotations();
excludeAnnotations(annotationsToExclude);
initializeDBs();
}
// use this constructor if you want to select specific annotations (and/or interfaces)
public VariantAnnotatorEngine(List<String> annotationGroupsToUse, List<String> annotationsToUse, AnnotatorCompatibleWalker walker, GenomeAnalysisEngine toolkit) {
public VariantAnnotatorEngine(List<String> annotationGroupsToUse, List<String> annotationsToUse, List<String> annotationsToExclude, AnnotatorCompatibleWalker walker, GenomeAnalysisEngine toolkit) {
this.walker = walker;
this.toolkit = toolkit;
initializeAnnotations(annotationGroupsToUse, annotationsToUse);
initializeAnnotations(annotationGroupsToUse, annotationsToUse, annotationsToExclude);
initializeDBs();
}
@ -96,10 +97,30 @@ public class VariantAnnotatorEngine {
requestedExpressions.add(new VAExpression(expression, walker.getResourceRodBindings()));
}
private void initializeAnnotations(List<String> annotationGroupsToUse, List<String> annotationsToUse) {
private void initializeAnnotations(List<String> annotationGroupsToUse, List<String> annotationsToUse, List<String> annotationsToExclude) {
AnnotationInterfaceManager.validateAnnotations(annotationGroupsToUse, annotationsToUse);
requestedInfoAnnotations = AnnotationInterfaceManager.createInfoFieldAnnotations(annotationGroupsToUse, annotationsToUse);
requestedGenotypeAnnotations = AnnotationInterfaceManager.createGenotypeAnnotations(annotationGroupsToUse, annotationsToUse);
excludeAnnotations(annotationsToExclude);
}
private void excludeAnnotations(List<String> annotationsToExclude) {
if ( annotationsToExclude.size() == 0 )
return;
List<InfoFieldAnnotation> tempRequestedInfoAnnotations = new ArrayList<InfoFieldAnnotation>(requestedInfoAnnotations.size());
for ( InfoFieldAnnotation annotation : requestedInfoAnnotations ) {
if ( !annotationsToExclude.contains(annotation.getClass().getSimpleName()) )
tempRequestedInfoAnnotations.add(annotation);
}
requestedInfoAnnotations = tempRequestedInfoAnnotations;
List<GenotypeAnnotation> tempRequestedGenotypeAnnotations = new ArrayList<GenotypeAnnotation>(requestedGenotypeAnnotations.size());
for ( GenotypeAnnotation annotation : requestedGenotypeAnnotations ) {
if ( !annotationsToExclude.contains(annotation.getClass().getSimpleName()) )
tempRequestedGenotypeAnnotations.add(annotation);
}
requestedGenotypeAnnotations = tempRequestedGenotypeAnnotations;
}
private void initializeDBs() {

9
public/java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java
View File

@ -149,6 +149,13 @@ public class UnifiedGenotyper extends LocusWalker<VariantCallContext, UnifiedGen
@Argument(fullName="annotation", shortName="A", doc="One or more specific annotations to apply to variant calls", required=false)
protected List<String> annotationsToUse = new ArrayList<String>();
/**
* Which annotations to exclude from output in the VCF file. Note that this argument has higher priority than the -A or -G arguments,
* so annotations will be excluded even if they are explicitly included with the other options.
*/
@Argument(fullName="excludeAnnotation", shortName="XA", doc="One or more specific annotations to exclude", required=false)
protected List<String> annotationsToExclude = new ArrayList<String>();
/**
* Which groups of annotations to add to the output VCF file. See the VariantAnnotator -list argument to view available groups.
*/
@ -210,7 +217,7 @@ public class UnifiedGenotyper extends LocusWalker<VariantCallContext, UnifiedGen
if ( verboseWriter != null )
verboseWriter.println("AFINFO\tLOC\tREF\tALT\tMAF\tF\tAFprior\tAFposterior\tNormalizedPosterior");
annotationEngine = new VariantAnnotatorEngine(Arrays.asList(annotationClassesToUse), annotationsToUse, this, getToolkit());
annotationEngine = new VariantAnnotatorEngine(Arrays.asList(annotationClassesToUse), annotationsToUse, annotationsToExclude, this, getToolkit());
UG_engine = new UnifiedGenotyperEngine(getToolkit(), UAC, logger, verboseWriter, annotationEngine, samples);
// initialize the header

26
public/java/test/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotatorIntegrationTest.java
View File

@ -31,7 +31,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testHasAnnotsAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample2.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
baseTestString() + " -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample2.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("8e7de435105499cd71ffc099e268a83e"));
executeTest("test file has annotations, asking for annotations, #1", spec);
}
@ -39,7 +39,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testHasAnnotsAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample3.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,050,000", 1,
baseTestString() + " -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample3.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,050,000", 1,
Arrays.asList("64b6804cb1e27826e3a47089349be581"));
executeTest("test file has annotations, asking for annotations, #2", spec);
}
@ -63,7 +63,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testNoAnnotsAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample2empty.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
baseTestString() + " -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample2empty.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("fd1ffb669800c2e07df1e2719aa38e49"));
executeTest("test file doesn't have annotations, asking for annotations, #1", spec);
}
@ -71,15 +71,23 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testNoAnnotsAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,050,000", 1,
baseTestString() + " -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,050,000", 1,
Arrays.asList("09f8e840770a9411ff77508e0ed0837f"));
executeTest("test file doesn't have annotations, asking for annotations, #2", spec);
}
@Test
public void testExcludeAnnotations() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G Standard -XA FisherStrand -XA ReadPosRankSumTest --variant:VCF3 " + validationDataLocation + "vcfexample2empty.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("b49fe03aa4b675db80a9db38a3552c95"));
executeTest("test exclude annotations", spec);
}
@Test
public void testOverwritingHeader() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF " + validationDataLocation + "vcfexample4.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,001,292", 1,
baseTestString() + " -G Standard --variant:VCF " + validationDataLocation + "vcfexample4.vcf -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,001,292", 1,
Arrays.asList("78d2c19f8107d865970dbaf3e12edd92"));
executeTest("test overwriting header", spec);
}
@ -87,7 +95,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testNoReads() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
baseTestString() + " -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
Arrays.asList("16e3a1403fc376320d7c69492cad9345"));
executeTest("not passing it any reads", spec);
}
@ -103,7 +111,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testDBTagWithHapMap() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " --comp:H3 " + validationDataLocation + "fakeHM3.vcf -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
baseTestString() + " --comp:H3 " + validationDataLocation + "fakeHM3.vcf -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
Arrays.asList("1bc01c5b3bd0b7aef75230310c3ce688"));
executeTest("getting DB tag with HM3", spec);
}
@ -111,7 +119,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
@Test
public void testUsingExpression() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " --resource:foo " + validationDataLocation + "targetAnnotations.vcf -G \"Standard\" --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -E foo.AF -BTI variant", 1,
baseTestString() + " --resource:foo " + validationDataLocation + "targetAnnotations.vcf -G Standard --variant:VCF3 " + validationDataLocation + "vcfexample3empty.vcf -E foo.AF -BTI variant", 1,
Arrays.asList("e9c0d832dc6b4ed06c955060f830c140"));
executeTest("using expression", spec);
}
@ -121,7 +129,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
final String MD5 = "13269d5a2e16f06fd755cc0fb9271acf";
for ( String file : Arrays.asList("CEU.exon.2010_03.sites.vcf", "CEU.exon.2010_03.sites.vcf.gz")) {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -A HomopolymerRun --variant:VCF " + validationDataLocation + "/" + file + " -BTI variant -NO_HEADER", 1,
baseTestString() + " -A HomopolymerRun --variant:VCF " + validationDataLocation + file + " -BTI variant -NO_HEADER", 1,
Arrays.asList(MD5));
executeTest("Testing lookup vcf tabix vs. vcf tribble", spec);
}