Merge pull request #1135 from broadinstitute/sl_issue_1056

Added file-extension--dependent interval-list output to RealignerTargetCreator.

protected/gatk-tools-protected/src/main/java/org/broadinstitute/gatk/tools/walkers/indels/RealignerTargetCreator.java

@@ -51,6 +51,11 @@
 
 package org.broadinstitute.gatk.tools.walkers.indels;
 
+import htsjdk.samtools.SAMFileHeader;
+import htsjdk.samtools.util.IOUtil;
+import htsjdk.samtools.util.Interval;
+import htsjdk.samtools.util.IntervalList;
+import org.apache.commons.io.FilenameUtils;
 import org.broadinstitute.gatk.engine.walkers.*;
 import org.broadinstitute.gatk.utils.commandline.Argument;
 import org.broadinstitute.gatk.utils.commandline.Input;
@@ -61,6 +66,7 @@ import org.broadinstitute.gatk.utils.contexts.AlignmentContext;
 import org.broadinstitute.gatk.utils.contexts.ReferenceContext;
 import org.broadinstitute.gatk.engine.filters.*;
 import org.broadinstitute.gatk.engine.iterators.ReadTransformer;
+import org.broadinstitute.gatk.utils.exceptions.GATKException;
 import org.broadinstitute.gatk.utils.refdata.RefMetaDataTracker;
 import org.broadinstitute.gatk.utils.GenomeLoc;
 import org.broadinstitute.gatk.utils.exceptions.UserException;
@@ -70,7 +76,9 @@ import org.broadinstitute.gatk.utils.pileup.PileupElement;
 import org.broadinstitute.gatk.utils.pileup.ReadBackedPileup;
 import htsjdk.variant.variantcontext.VariantContext;
 
-import java.io.PrintStream;
+import java.io.BufferedWriter;
+import java.io.File;
+import java.io.IOException;
 import java.util.ArrayList;
 import java.util.Collections;
 import java.util.List;
@@ -143,7 +151,7 @@ public class RealignerTargetCreator extends RodWalker<RealignerTargetCreator.Eve
      * The target intervals for realignment.
      */
     @Output
-    protected PrintStream out;
+    protected File out;
 
     /**
      * Any number of VCF files representing known SNPs and/or indels.  Could be e.g. dbSNP and/or official 1000 Genomes indel calls.
@@ -276,8 +284,24 @@ public class RealignerTargetCreator extends RodWalker<RealignerTargetCreator.Eve
         if ( sum.right != null && sum.right.isReportableEvent() )
             sum.intervals.add(sum.right.getLoc());
 
-        for ( GenomeLoc loc : sum.intervals )
+        if ( FilenameUtils.getExtension(out.getName()).equals("interval_list") ) {
-            out.println(loc);
+            final SAMFileHeader masterSequenceDictionaryHeader = new SAMFileHeader();
+            masterSequenceDictionaryHeader.setSequenceDictionary(getToolkit().getMasterSequenceDictionary());
+            final IntervalList intervalList = new IntervalList(masterSequenceDictionaryHeader);
+            for ( GenomeLoc loc : sum.intervals ) {
+                intervalList.add(new Interval(loc.getContig(), loc.getStart(), loc.getStop()));
+            }
+            intervalList.write(out);
+        } else {
+            try ( BufferedWriter bufferedWriter = IOUtil.openFileForBufferedWriting(out) ) {
+                for ( GenomeLoc loc : sum.intervals ) {
+                    bufferedWriter.write(loc.toString());
+                    bufferedWriter.newLine();
+                }
+            } catch (final IOException e) {
+                throw new GATKException("Error writing out intervals to file: " + out.getAbsolutePath(), e);
+            }
+        }
     }
 
     public EventPair reduceInit() {

protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/indels/RealignerTargetCreatorIntegrationTest.java

@@ -51,45 +51,60 @@
 
 package org.broadinstitute.gatk.tools.walkers.indels;
 
+import htsjdk.samtools.reference.ReferenceSequenceFile;
+import htsjdk.samtools.util.Interval;
+import htsjdk.samtools.util.IntervalList;
 import org.broadinstitute.gatk.engine.walkers.WalkerTest;
+import org.broadinstitute.gatk.utils.BaseTest;
+import org.broadinstitute.gatk.utils.GenomeLoc;
+import org.broadinstitute.gatk.utils.GenomeLocParser;
+import org.broadinstitute.gatk.utils.fasta.CachingIndexedFastaSequenceFile;
+import org.broadinstitute.gatk.utils.interval.IntervalUtils;
+import org.testng.Assert;
+import org.testng.annotations.DataProvider;
 import org.testng.annotations.Test;
 
+import java.io.File;
+import java.io.IOException;
+import java.util.ArrayList;
 import java.util.Arrays;
+import java.util.Collections;
+import java.util.List;
 
 public class RealignerTargetCreatorIntegrationTest extends WalkerTest {
 
-    @Test
+    @DataProvider(name = "intervals1")
-    public void testIntervals1() {
+    public Object[][] intervals1() {
-        String md5 = "3f0b63a393104d0c4158c7d1538153b8";
+        String arguments = "-T RealignerTargetCreator -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam --mismatchFraction 0.15 -L 1:10,000,000-10,050,000";
-
+        return new Object[][]{
-        WalkerTest.WalkerTestSpec spec1 = new WalkerTest.WalkerTestSpec(
+                {"test standard nt=1", arguments},
-                "-T RealignerTargetCreator -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam --mismatchFraction 0.15 -L 1:10,000,000-10,050,000 -o %s",
+                {"test standard nt=4", "-nt 4 " + arguments}
-                 1,
+        };
-                 Arrays.asList(md5));
-        executeTest("test standard nt=1", spec1);
-
-        WalkerTest.WalkerTestSpec spec2 = new WalkerTest.WalkerTestSpec(
-                "-nt 4 -T RealignerTargetCreator -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam --mismatchFraction 0.15 -L 1:10,000,000-10,050,000 -o %s",
-                 1,
-                 Arrays.asList(md5));
-        executeTest("test standard nt=4", spec2);
     }
 
-    @Test
+    @DataProvider(name = "intervals2")
-    public void testIntervals2() {
+    public Object[][] intervals2() {
+        String arguments = "-T RealignerTargetCreator --known " + b36dbSNP129 + " -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,200,000";
+        return new Object[][]{
+                {"test with dbsnp nt=1", arguments},
+                {"test with dbsnp nt=4", "-nt 4 " + arguments}
+        };
+    }
+
+    @Test(dataProvider = "intervals1")
+    public void testIntervals1(String testName, String arguments) {
+        String md5 = "3f0b63a393104d0c4158c7d1538153b8";
+
+        WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(arguments + " -o %s", 1, Arrays.asList(md5));
+        executeTest(testName, spec);
+    }
+
+    @Test(dataProvider = "intervals2")
+    public void testIntervals2(String testName, String arguments) {
         String md5 = "d073237694175c75d37bd4f40b8c64db";
 
-        WalkerTest.WalkerTestSpec spec1 = new WalkerTest.WalkerTestSpec(
+        WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(arguments + " -o %s", 1, Arrays.asList(md5));
-                "-T RealignerTargetCreator --known " + b36dbSNP129 + " -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,200,000 -o %s",
+        executeTest(testName, spec);
-                 1,
-                 Arrays.asList(md5));
-        executeTest("test with dbsnp nt=1", spec1);
-
-        WalkerTest.WalkerTestSpec spec2 = new WalkerTest.WalkerTestSpec(
-                "-nt 4 -T RealignerTargetCreator --known " + b36dbSNP129 + " -R " + b36KGReference + " -I " + validationDataLocation + "NA12878.1kg.p2.chr1_10mb_11_mb.SLX.bam -L 1:10,000,000-10,200,000 -o %s",
-                 1,
-                 Arrays.asList(md5));
-        executeTest("test with dbsnp nt=4", spec2);
     }
 
     @Test
@@ -110,4 +125,35 @@ public class RealignerTargetCreatorIntegrationTest extends WalkerTest {
                 Arrays.asList(""));
         executeTest("test bad cigar string string does not fail", spec);
     }
+
+    @Test(dataProvider = "intervals1")
+    public void testTargetListAgainstIntervalList(String testName, String arguments) throws IOException {
+        final List<String> md5 = Collections.emptyList();
+        final File targetListFile = createTempFile("RTCTest", ".targets");
+        final File intervalListFile = createTempFile("RTCTest", ".interval_list");
+
+        WalkerTest.WalkerTestSpec targetListSpec = new WalkerTest.WalkerTestSpec(arguments, 1, md5);
+        WalkerTest.WalkerTestSpec intervalListSpec = new WalkerTest.WalkerTestSpec(arguments, 1, md5);
+
+        targetListSpec.setOutputFileLocation(targetListFile);
+        intervalListSpec.setOutputFileLocation(intervalListFile);
+
+        executeTest(testName + " (compare target-list and interval-list output)", targetListSpec);
+        executeTest(testName + " (compare target-list and interval-list output)", intervalListSpec);
+
+        final ReferenceSequenceFile seq = new CachingIndexedFastaSequenceFile(new File(BaseTest.hg19Reference));
+        final GenomeLocParser hg19GenomeLocParser = new GenomeLocParser(seq);
+        final List<GenomeLoc> targetList = IntervalUtils.intervalFileToList(hg19GenomeLocParser,
+                targetListFile.getAbsolutePath());
+        final List<Interval> targetListResult = new ArrayList<>();
+        for ( GenomeLoc target : targetList ) {
+            targetListResult.add(new Interval(target.getContig(), target.getStart(), target.getStop()));
+        }
+
+        final List<Interval> intervalListResult = IntervalList.fromFile(intervalListFile).getIntervals();
+
+        Assert.assertFalse(targetListResult.isEmpty());
+        Assert.assertFalse(intervalListResult.isEmpty());
+        Assert.assertEquals(targetListResult, intervalListResult);
+    }
 }