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

adding changes for VCF 4, mostly in the way we handle VCF headers. The header fields are now aware of the differences between different VCF formats. There was also a bunch of clean-up of out-of-spec VCF used in the tests (mismatched VCF file format fields, etc), and updates to the associated integration tests. Also some logging statements for BTI. 

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@3584 348d0f76-0448-11de-a6fe-93d51630548a
This commit is contained in:
aaron 2010-06-18 08:23:23 +00:00
parent 32f6781ac7
commit b978d5946b
12 changed files with 38 additions and 41 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
java/src/org/broadinstitute/sting/gatk/GenomeAnalysisEngine.java
View File

@ -252,6 +252,7 @@ public class GenomeAnalysisEngine {
for (String str : rodNames.keySet())
if (str.equals(rodName)) {
logger.info("Adding interval list from track (ROD) named " + rodName);
RMDIntervalGenerator intervalGenerator = new RMDIntervalGenerator(rodNames.get(str).getReferenceOrderedData());
ret.addAll(intervalGenerator.toGenomeLocList());
}

7
java/src/org/broadinstitute/sting/gatk/walkers/genotyper/UnifiedGenotyper.java
View File

@ -25,10 +25,7 @@
package org.broadinstitute.sting.gatk.walkers.genotyper;
import org.broad.tribble.vcf.VCFGenotypeRecord;
import org.broad.tribble.vcf.VCFHeaderLine;
import org.broad.tribble.vcf.VCFInfoHeaderLine;
import org.broad.tribble.vcf.VCFRecord;
import org.broad.tribble.vcf.*;
import org.broadinstitute.sting.gatk.GenomeAnalysisEngine;
import org.broadinstitute.sting.gatk.contexts.*;
import org.broadinstitute.sting.gatk.refdata.RefMetaDataTracker;
@ -156,7 +153,7 @@ public class UnifiedGenotyper extends LocusWalker<VariantCallContext, UnifiedGen
headerInfo.add(new VCFInfoHeaderLine(VCFRecord.STRAND_BIAS_KEY, 1, VCFInfoHeaderLine.INFO_TYPE.Float, "Strand Bias"));
// FORMAT and INFO fields
headerInfo.addAll(VCFGenotypeRecord.getSupportedHeaderStrings());
headerInfo.addAll(VCFGenotypeRecord.getSupportedHeaderStrings(VCFHeaderVersion.VCF3_3));
// all of the arguments from the argument collection
Set<Object> args = new HashSet<Object>();

3
java/src/org/broadinstitute/sting/utils/genotype/GenotypeWriterFactory.java
View File

@ -3,6 +3,7 @@ package org.broadinstitute.sting.utils.genotype;
import net.sf.samtools.SAMFileHeader;
import org.broad.tribble.vcf.VCFHeaderLine;
import org.broad.tribble.vcf.VCFGenotypeRecord;
import org.broad.tribble.vcf.VCFHeaderVersion;
import org.broadinstitute.sting.utils.StingException;
import org.broadinstitute.sting.utils.genotype.geli.*;
import org.broadinstitute.sting.utils.genotype.glf.*;
@ -68,7 +69,7 @@ public class GenotypeWriterFactory {
// VCF
if ( writer instanceof VCFGenotypeWriter ) {
if ( headerInfo == null )
headerInfo = new HashSet<VCFHeaderLine>(VCFGenotypeRecord.getSupportedHeaderStrings());
headerInfo = new HashSet<VCFHeaderLine>(VCFGenotypeRecord.getSupportedHeaderStrings(VCFHeaderVersion.VCF3_3));
((VCFGenotypeWriter)writer).writeHeader(sampleNames, headerInfo);
}
// GELI

7
java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFGenotypeWriterAdapter.java
View File

@ -1,9 +1,6 @@
package org.broadinstitute.sting.utils.genotype.vcf;
import org.broad.tribble.vcf.VCFFormatHeaderLine;
import org.broad.tribble.vcf.VCFHeader;
import org.broad.tribble.vcf.VCFHeaderLine;
import org.broad.tribble.vcf.VCFRecord;
import org.broad.tribble.vcf.*;
import org.broadinstitute.sting.gatk.contexts.variantcontext.VariantContext;
import org.broadinstitute.sting.gatk.refdata.VariantContextAdaptors;
import org.apache.log4j.Logger;
@ -57,7 +54,7 @@ public class VCFGenotypeWriterAdapter implements VCFGenotypeWriter {
// set up the header fields
Set<VCFHeaderLine> hInfo = new TreeSet<VCFHeaderLine>();
hInfo.add(new VCFHeaderLine(VCFHeader.FILE_FORMAT_KEY, VCFHeader.VCF_VERSION));
hInfo.add(new VCFHeaderLine(VCFHeaderVersion.VCF3_3.getFormatString(), VCFHeaderVersion.VCF3_3.getVersionString()));
// set up the allowed genotype format fields
if ( headerInfo != null ) {

7
java/src/org/broadinstitute/sting/utils/genotype/vcf/VCFWriter.java
View File

@ -3,6 +3,7 @@ package org.broadinstitute.sting.utils.genotype.vcf;
import org.broad.tribble.vcf.VCFHeader;
import org.broad.tribble.vcf.VCFHeaderLine;
import org.broad.tribble.vcf.VCFHeaderVersion;
import org.broad.tribble.vcf.VCFRecord;
import java.io.*;
@ -52,11 +53,11 @@ public class VCFWriter {
// the fileformat field needs to be written first
TreeSet<VCFHeaderLine> nonFormatMetaData = new TreeSet<VCFHeaderLine>();
for ( VCFHeaderLine line : header.getMetaData() ) {
if ( line.getKey().equals(VCFHeader.FILE_FORMAT_KEY) ) {
if ( line.getKey().equals(VCFHeaderVersion.VCF3_3.getFormatString()) ) {
mWriter.write(VCFHeader.METADATA_INDICATOR + line.toString() + "\n");
}
else if ( line.getKey().equals(VCFHeader.OLD_FILE_FORMAT_KEY) ) {
mWriter.write(VCFHeader.METADATA_INDICATOR + VCFHeader.FILE_FORMAT_KEY + line.toString().substring(VCFHeader.OLD_FILE_FORMAT_KEY.length()) + "\n");
else if ( line.getKey().equals(VCFHeaderVersion.VCF3_2.getFormatString()) ) {
mWriter.write(VCFHeader.METADATA_INDICATOR + VCFHeaderVersion.VCF3_2.getFormatString() + "=" + VCFHeaderVersion.VCF3_2.getVersionString() + "\n");
} else {
nonFormatMetaData.add(line);
}

2
java/test/org/broadinstitute/sting/gatk/walkers/VariantsToVCFIntegrationTest.java
View File

@ -70,7 +70,7 @@ public class VariantsToVCFIntegrationTest extends WalkerTest {
@Test
public void testGenotypesToVCFUsingVCFInput() {
List<String> md5 = new ArrayList<String>();
md5.add("92d661a3789e55078197666eb9ee7020");
md5.add("3f920c6a443764b183e4765b4e4d00b0");
WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(
"-R " + oneKGLocation + "reference/human_b36_both.fasta" +

2
java/test/org/broadinstitute/sting/gatk/walkers/annotator/SecondBaseSkewIntegrationTest.java
View File

@ -63,7 +63,7 @@ public class SecondBaseSkewIntegrationTest extends WalkerTest {
+ " -R " + seqLocation + "references/Homo_sapiens_assembly18/v0/Homo_sapiens_assembly18.fasta -A SecondBaseSkew"
+ " -sample variant -B variant,VCF," + validationDataLocation + "FHS_pileup_test_chr15.vcf"
+ " -o %s -L chr15:46347148";
String expected_md5 = "465f92e689110afeb308c201cb6e8c5a";
String expected_md5 = "f4aff94a713fcf4dc5750d6a6a884ad3";
WalkerTestSpec spec = new WalkerTestSpec(test_args,1,Arrays.asList(expected_md5));
executeTest("Testing on locus with many indels", spec);
}

22
java/test/org/broadinstitute/sting/gatk/walkers/annotator/VariantAnnotatorIntegrationTest.java
View File

@ -30,7 +30,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
case 2: return "-L chr22:20660081-20660083 -L chr22:29198100-29198104 -L chr22:29821330-29821334";
default: throw new StingException("Impossible test has been run: secondbasetest number "+testNo);
}
}
}
public static String secondBaseTestmd5( int testNo ) {
switch ( testNo ) {
@ -50,7 +50,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testHasAnnotsNotAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("5a91f6b50fc136d7d3f8735dbc64defe"));
Arrays.asList("92369e62b3bade625992bc1741e9367a"));
executeTest("test file has annotations, not asking for annotations, #1", spec);
}
@ -58,7 +58,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testHasAnnotsNotAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B variant,VCF," + 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("0a561b161a06e68b88417ff5fe365871"));
Arrays.asList("1a394cdace4ff7698b603c2252ee9365"));
executeTest("test file has annotations, not asking for annotations, #2", spec);
}
@ -66,7 +66,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testHasAnnotsAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("dff7c888fbd142b6eff27c3233c1292d"));
Arrays.asList("5a8c9c18da89b052e7f3797cd391bff6"));
executeTest("test file has annotations, asking for annotations, #1", spec);
}
@ -74,7 +74,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testHasAnnotsAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" -B variant,VCF," + 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("45c9d5136900d9ab347f489d6f442bb4"));
Arrays.asList("0c8607c69d761bd058b944d1e98708b6"));
executeTest("test file has annotations, asking for annotations, #2", spec);
}
@ -82,7 +82,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testNoAnnotsNotAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B variant,VCF," + validationDataLocation + "vcfexample2empty.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("16e578597eed68609268c00886f2842a"));
Arrays.asList("006642aabd41e806d3ea64bf5441f1d2"));
executeTest("test file doesn't have annotations, not asking for annotations, #1", spec);
}
@ -90,7 +90,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testNoAnnotsNotAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B variant,VCF," + 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("0e947087c44479faefe616f6b6f7d272"));
Arrays.asList("c66ed83d2a981c43546d05120809ed57"));
executeTest("test file doesn't have annotations, not asking for annotations, #2", spec);
}
@ -98,7 +98,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testNoAnnotsAsking1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" -B variant,VCF," + validationDataLocation + "vcfexample2empty.vcf -I " + validationDataLocation + "low_coverage_CEU.chr1.10k-11k.bam -L 1:10,020,000-10,021,000", 1,
Arrays.asList("7c13af3226a92bfc3826c37b30f179e0"));
Arrays.asList("3b570e13bd0e1cb5750da268178ce656"));
executeTest("test file doesn't have annotations, asking for annotations, #1", spec);
}
@ -106,7 +106,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testNoAnnotsAsking2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" -B variant,VCF," + 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("369ad538eacdc0943b58a807d92d823c"));
Arrays.asList("42e9c57cd5d266fd7837354fc8495176"));
executeTest("test file doesn't have annotations, asking for annotations, #2", spec);
}
@ -114,7 +114,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testNoReads() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -G \"Standard\" -B variant,VCF," + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
Arrays.asList("07af9983127c62e96accc03db2fb523e"));
Arrays.asList("374da32e1b9540bfe8ee6ea5b896babd"));
executeTest("not passing it any reads", spec);
}
@ -122,7 +122,7 @@ public class VariantAnnotatorIntegrationTest extends WalkerTest {
public void testDBTag() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -D " + GATKDataLocation + "dbsnp_129_b36.rod -G \"Standard\" -B variant,VCF," + validationDataLocation + "vcfexample3empty.vcf -BTI variant", 1,
Arrays.asList("286887826ab99bb2864f7f4db195e59e"));
Arrays.asList("05a9dcb87efc65e5c726c201460192d9"));
executeTest("getting DB tag", spec);
}
}

10
java/test/org/broadinstitute/sting/gatk/walkers/concordance/CallsetConcordanceIntegrationTest.java
View File

@ -14,7 +14,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
public void testSimpleVenn() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B set1,VCF," + validationDataLocation + "NA12878.example1.vcf -B set2,VCF," + validationDataLocation + "NA12878.example2.vcf -CT SimpleVenn", 1,
Arrays.asList("d9124d2b0fb5bec5bc50c26a16b4e900"));
Arrays.asList("a1970effe9c51923d52af9034e778de4"));
executeTest("testSimpleVenn", spec);
}
@ -22,7 +22,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
public void testSNPConcordance() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B set1,VCF," + validationDataLocation + "NA12878.example1.vcf -B set2,VCF," + validationDataLocation + "NA12878.example2.vcf -CT SNPGenotypeConcordance:qscore=5", 1,
Arrays.asList("df1fbc744947f316f65f51a21368b0e4"));
Arrays.asList("e7a0d52c266ba3c76283111674c7168f"));
executeTest("testSNPConcordance", spec);
}
@ -30,7 +30,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
public void testNWayVenn() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B set1,VCF," + validationDataLocation + "NA12878.example1.vcf -B set2,VCF," + validationDataLocation + "NA12878.example2.vcf -B set3,VCF," + validationDataLocation + "CEU.sample.vcf -CT NWayVenn", 1,
Arrays.asList("aa835ae5368b35f376b844d7f8ef2976"));
Arrays.asList("e65fc811137fca7d6c32125240c7468f"));
executeTest("testNWayVenn", spec);
}
@ -38,7 +38,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
public void testMulti() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B set1,VCF," + validationDataLocation + "NA12878.example1.vcf -B set2,VCF," + validationDataLocation + "NA12878.example2.vcf -CT SimpleVenn -CT NWayVenn -CT SNPGenotypeConcordance:qscore=5", 1,
Arrays.asList("c9ef68cc3b7dc08f1d2b49170e6560ab"));
Arrays.asList("ddc2507590e28743e9cb4b132cb066e7"));
executeTest("testMulti", spec);
}
@ -46,7 +46,7 @@ public class CallsetConcordanceIntegrationTest extends WalkerTest {
public void testComplex() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B set1,VCF," + validationDataLocation + "complexExample.vcf -B set2,VCF," + validationDataLocation + "complexExample.vcf -CT NWayVenn", 1,
Arrays.asList("7bc72ec5f8b0fda5d59ebd2526b53e48"));
Arrays.asList("250df7bde7a8cf9c7ee7c5704183ea88"));
executeTest("testComplex", spec);
}
}

12
java/test/org/broadinstitute/sting/gatk/walkers/filters/VariantFiltrationIntegrationTest.java
View File

@ -16,7 +16,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testNoAction() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("af14c7d03e3516b61f2702c9e4a7780f"));
Arrays.asList("d2f0a3c6e598d9054de9ea7bd914a12d"));
executeTest("test no action", spec);
}
@ -24,7 +24,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testClusteredSnps() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -window 10 -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("b2b18929289bb47f07bcc23d4cec94c4"));
Arrays.asList("4dad2ddfb74785c5c80c2a73709e543a"));
executeTest("test clustered SNPs", spec);
}
@ -32,7 +32,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testMask() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -mask foo -B mask,VCF," + validationDataLocation + "vcfexample2.vcf -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("af048053eaef84fc4d61c51c50be1e0a"));
Arrays.asList("70fc964e73fda55ed3f2c0b0ae4bcec1"));
executeTest("test mask", spec);
}
@ -40,7 +40,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testFilter1() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -filter 'DoC < 20 || FisherStrand > 20.0' -filterName foo -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("43a580d9c684a496f21d0d42939dd910"));
Arrays.asList("bde496530822a81197df31f5ecf98033"));
executeTest("test filter #1", spec);
}
@ -48,7 +48,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testFilter2() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " -filter 'AlleleBalance < 70.0 && FisherStrand == 1.4' -filterName bar -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("2bd1e8975d8105dec2fb6055fbf00569"));
Arrays.asList("750f5076e906901dabec78f804f035fa"));
executeTest("test filter #2", spec);
}
@ -56,7 +56,7 @@ public class VariantFiltrationIntegrationTest extends WalkerTest {
public void testFilterWithSeparateNames() {
WalkerTestSpec spec = new WalkerTestSpec(
baseTestString() + " --filterName ABF -filter 'AlleleBalance < 70.0' --filterName FSF -filter 'FisherStrand == 1.4' -B variant,VCF," + validationDataLocation + "vcfexample2.vcf -L 1:10,020,000-10,021,000", 1,
Arrays.asList("3479158ffd02a45371b5103277a30a53"));
Arrays.asList("7f876c9187948d27c0cd39f5a0395e0e"));
executeTest("test filter with separate names #2", spec);
}
}

2
java/test/org/broadinstitute/sting/gatk/walkers/variantrecalibration/VariantRecalibrationWalkersIntegrationTest.java
View File

@ -39,7 +39,7 @@ public class VariantRecalibrationWalkersIntegrationTest extends WalkerTest {
@Test
public void testVariantRecalibrator() {
HashMap<String, String> e = new HashMap<String, String>();
e.put( validationDataLocation + "yri.trio.gatk_glftrio.intersection.annotated.filtered.chr1.vcf", "412bdb2eb4ca8f7ee9dfb39cda676c95" );
e.put( validationDataLocation + "yri.trio.gatk_glftrio.intersection.annotated.filtered.chr1.vcf", "cf8dac86df97d55f866dd379434bbdc2" );
for ( Map.Entry<String, String> entry : e.entrySet() ) {
String vcf = entry.getKey();

4
java/test/org/broadinstitute/sting/utils/genotype/vcf/VCFWriterUnitTest.java
View File

@ -66,8 +66,8 @@ public class VCFWriterUnitTest extends BaseTest {
* @return a fake VCF header
*/
public static VCFHeader createFakeHeader(Set<VCFHeaderLine> metaData, Set<String> additionalColumns) {
metaData.add(new VCFHeaderLine(VCFHeader.FILE_FORMAT_KEY, VCFHeader.VCF_VERSION));
metaData.add(new VCFHeaderLine("two", "2"));
metaData.add(new VCFHeaderLine(VCFHeaderVersion.VCF3_3.getFormatString(), VCFHeaderVersion.VCF3_3.getVersionString(),VCFHeaderVersion.VCF3_3));
metaData.add(new VCFHeaderLine("two", "2",VCFHeaderVersion.VCF3_3));
additionalColumns.add("FORMAT");
additionalColumns.add("extra1");
additionalColumns.add("extra2");