Merge pull request #1331 from broadinstitute/rhl_doc_null_ptr

Add informative exceptions to getSAMFileSamples()
broadgsa · Mar 31, 2016 · edbb7e7 · edbb7e7
2 parents 68b068d + e4003bc
commit edbb7e7
Show file tree

Hide file tree

Showing 4 changed files with 56 additions and 4 deletions.
diff --git a/...t/java/org/broadinstitute/gatk/tools/walkers/coverage/DepthOfCoverageIntegrationTest.java b/...t/java/org/broadinstitute/gatk/tools/walkers/coverage/DepthOfCoverageIntegrationTest.java
@@ -235,4 +235,24 @@ public void testSampleGeneSummaryHeaderFix(){
 
         execute("testSampleGeneSummaryHeaderFix", spec);
     }
+
+    @Test(expectedExceptions = RuntimeException.class)
+    public void testMissingSAMHeaderReadGroup() {
+        final String[] intervals = {"chr1:200-300"};
+        final String[] bams = {privateTestDir + "exampleBAMNoRG.bam"};
+
+        final String cmd = buildRootCmd(exampleFASTA,new ArrayList<>(Arrays.asList(bams)),new ArrayList<>(Arrays.asList(intervals)));
+        final WalkerTestSpec spec = new WalkerTestSpec(cmd, 0, new ArrayList<String>());
+        execute("testMissingSAMHeaderReadGroup", spec);
+    }
+
+    @Test(expectedExceptions = RuntimeException.class)
+    public void testMissingSAMHeaderReadGroupSample() {
+        final String[] intervals = {"chr1:200-300"};
+        final String[] bams = {privateTestDir + "exampleBAMNoSM.bam"};
+
+        final String cmd = buildRootCmd(exampleFASTA,new ArrayList<>(Arrays.asList(bams)),new ArrayList<>(Arrays.asList(intervals)));
+        final WalkerTestSpec spec = new WalkerTestSpec(cmd, 0, new ArrayList<String>());
+        execute("testMissingSAMHeaderReadGroupSample", spec);
+    }
 }
diff --git a/public/gatk-utils/src/main/java/org/broadinstitute/gatk/utils/help/HelpConstants.java b/public/gatk-utils/src/main/java/org/broadinstitute/gatk/utils/help/HelpConstants.java
@@ -27,7 +27,7 @@
 
 public class HelpConstants {
 
-    public final static String BASE_GATK_URL = "http://www.broadinstitute.org/gatk";
+    public final static String BASE_GATK_URL = "https://www.broadinstitute.org/gatk";
     public final static String GATK_DOCS_URL = BASE_GATK_URL + "/guide/tooldocs/";
     public final static String GATK_ARTICLE_URL = BASE_GATK_URL + "/guide/article";
     public final static String GATK_FORUM_URL = "http://gatkforums.broadinstitute.org/";

diff --git a/public/gatk-utils/src/main/java/org/broadinstitute/gatk/utils/sam/ReadUtils.java b/public/gatk-utils/src/main/java/org/broadinstitute/gatk/utils/sam/ReadUtils.java
@@ -32,8 +32,9 @@
 import org.broadinstitute.gatk.utils.*;
 import org.broadinstitute.gatk.utils.collections.Pair;
 import org.broadinstitute.gatk.utils.exceptions.ReviewedGATKException;
+import org.broadinstitute.gatk.utils.exceptions.UserException;
+import org.broadinstitute.gatk.utils.help.HelpConstants;
 
-import java.io.File;
 import java.util.*;
 
 /**
@@ -63,11 +64,26 @@ private ReadUtils() {
      * @return list of strings representing the sample names
      */
     public static Set<String> getSAMFileSamples(final SAMFileHeader header) {
+        if ( header == null ) {
+            throw new IllegalArgumentException("Missing SAM file header. " +
+                    "For more information on read groups, see " + HelpConstants.articlePost("6472"));
+        }
+
         // get all of the unique sample names
         final Set<String> samples = new TreeSet<String>();
-        List<SAMReadGroupRecord> readGroups = header.getReadGroups();
-        for ( SAMReadGroupRecord readGroup : readGroups )
+        final List<SAMReadGroupRecord> readGroups = header.getReadGroups();
+        if ( readGroups == null ) {
+            throw new UserException("SAM file header is missing the Read Group (@RG). " +
+                    "For more information on read groups, see " + HelpConstants.articlePost("6472"));
+        }
+        for ( final SAMReadGroupRecord readGroup : readGroups ) {
+            final String sample = readGroup.getSample();
+            if ( sample == null ) {
+                throw new UserException("SAM file header is missing the sample field (SM) in the Read Group (@RG). " +
+                        "For more information on read groups, see " + HelpConstants.articlePost("6472"));
+            }
             samples.add(readGroup.getSample());
+        }
         return samples;
     }
 

diff --git a/public/gatk-utils/src/test/java/org/broadinstitute/gatk/utils/sam/ReadUtilsUnitTest.java b/public/gatk-utils/src/test/java/org/broadinstitute/gatk/utils/sam/ReadUtilsUnitTest.java
@@ -25,11 +25,13 @@
 
 package org.broadinstitute.gatk.utils.sam;
 
+import htsjdk.samtools.SAMReadGroupRecord;
 import htsjdk.samtools.reference.IndexedFastaSequenceFile;
 import htsjdk.samtools.SAMFileHeader;
 import org.broadinstitute.gatk.utils.BaseTest;
 import org.broadinstitute.gatk.utils.BaseUtils;
 import org.broadinstitute.gatk.utils.Utils;
+import org.broadinstitute.gatk.utils.exceptions.UserException;
 import org.broadinstitute.gatk.utils.fasta.CachingIndexedFastaSequenceFile;
 import org.testng.Assert;
 import org.testng.annotations.DataProvider;
@@ -336,4 +338,18 @@ public Object[][] makeHasWellDefinedFragmentSizeData() throws Exception {
     private void testHasWellDefinedFragmentSize(final String name, final GATKSAMRecord read, final boolean expected) {
         Assert.assertEquals(ReadUtils.hasWellDefinedFragmentSize(read), expected);
     }
+
+    @Test(expectedExceptions = IllegalArgumentException.class)
+    public void testGetSAMFileMissingHeader() {
+        ReadUtils.getSAMFileSamples(null);
+    }
+
+    @Test(expectedExceptions = UserException.class)
+    public void testGetSAMFileMissingReadGroupsSamples() {
+        final SAMFileHeader header = ArtificialSAMUtils.createArtificialSamHeader();
+        final SAMReadGroupRecord samGroup = new SAMReadGroupRecord("id");
+        final List<SAMReadGroupRecord> list = new ArrayList<>(Arrays.asList(samGroup));
+        header.setReadGroups(list);
+        ReadUtils.getSAMFileSamples(header);
+    }
 }