From ca48ea78df9ddcd7cbebd1487bf0a3d3ed5c9211 Mon Sep 17 00:00:00 2001
From: hanna <hanna@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Tue, 7 Jun 2011 04:25:40 +0000
Subject: [PATCH] At Picard team's request, generate md5s for generated BAM
 files.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5954 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../gatk/io/storage/SAMFileWriterStorage.java |  7 +--
 .../SAMFileWriterArgumentTypeDescriptor.java  | 55 +++++++++++++++----
 .../gatk/io/stubs/SAMFileWriterStub.java      | 23 ++++++++
 3 files changed, 70 insertions(+), 15 deletions(-)

diff --git a/java/src/org/broadinstitute/sting/gatk/io/storage/SAMFileWriterStorage.java b/java/src/org/broadinstitute/sting/gatk/io/storage/SAMFileWriterStorage.java
index a3da3156c..610db1d76 100644
--- a/java/src/org/broadinstitute/sting/gatk/io/storage/SAMFileWriterStorage.java
+++ b/java/src/org/broadinstitute/sting/gatk/io/storage/SAMFileWriterStorage.java
@@ -60,16 +60,13 @@ public class SAMFileWriterStorage implements SAMFileWriter, Storage<SAMFileWrite
         // Enable automatic index creation for pre-sorted BAMs.
         if (stub.getFileHeader().getSortOrder().equals(SAMFileHeader.SortOrder.coordinate) && stub.getIndexOnTheFly())
             factory.setCreateIndex(true);
+        if (stub.getGenerateMD5())
+            factory.setCreateMd5File(true);
         // Adjust max records in RAM.
         if(stub.getMaxRecordsInRam() != null)
             factory.setMaxRecordsInRam(stub.getMaxRecordsInRam());
 
         if(stub.getSAMFile() != null) {
-            // HACK: Turn setCreateIndex off if the SAM file is null.  Picard has pledged to fix this
-            //       during the week of 14 Sept 2010.  Eliminate this check when they do.
-            if(stub.getSAMFile().getPath().equals("/dev/null"))
-                factory.setCreateIndex(false);
-
             try {
                 this.writer = createBAMWriter(factory,stub.getFileHeader(),stub.isPresorted(),file,stub.getCompressionLevel());
             }
diff --git a/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterArgumentTypeDescriptor.java b/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterArgumentTypeDescriptor.java
index bf2623f25..a9a272220 100644
--- a/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterArgumentTypeDescriptor.java
+++ b/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterArgumentTypeDescriptor.java
@@ -52,6 +52,7 @@ public class SAMFileWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor
     public static final String SIMPLIFY_BAM_SHORTNAME = SIMPLIFY_BAM_FULLNAME;
 
     public static final String DISABLE_INDEXING_FULLNAME = "disable_bam_indexing";
+    public static final String ENABLE_MD5_FULLNAME = "generate_md5";
 
     /**
      * The engine into which output stubs should be fed.
@@ -83,6 +84,7 @@ public class SAMFileWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor
         return Arrays.asList( createBAMArgumentDefinition(source),
                               createBAMCompressionArgumentDefinition(source),
                               disableWriteIndexArgumentDefinition(source),
+                              enableMD5GenerationArgumentDefinition(source),
                               createSimplifyBAMArgumentDefinition(source));
     }
 
@@ -102,23 +104,39 @@ public class SAMFileWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor
 
     @Override
     public Object parse( ParsingEngine parsingEngine, ArgumentSource source, Class type, ArgumentMatches matches )  {
+        // Extract all possible parameters that could be passed to a BAM file writer?
         ArgumentDefinition bamArgumentDefinition = createBAMArgumentDefinition(source);
         String writerFileName = getArgumentValue( bamArgumentDefinition, matches );
 
-        // This parser has been passed a null filename and the GATK is not responsible for creating a type default for the object;
-        // therefore, the user must have failed to specify a type default
-        if(writerFileName == null && !source.isRequired())
-            throw new MissingArgumentValueException(bamArgumentDefinition);
-
-        SAMFileWriterStub stub = new SAMFileWriterStub(engine, new File(writerFileName));
-
         String compressionLevelText = getArgumentValue( createBAMCompressionArgumentDefinition(source), matches );
         Integer compressionLevel = compressionLevelText != null ? Integer.valueOf(compressionLevelText) : null;
+
+        Boolean indexOnTheFly = !argumentIsPresent(disableWriteIndexArgumentDefinition(source),matches) ? true : null;
+        Boolean generateMD5 = argumentIsPresent(this.enableMD5GenerationArgumentDefinition(source),matches) ? true : null;
+        Boolean simplifyBAM = argumentIsPresent(createSimplifyBAMArgumentDefinition(source),matches);
+
+        // Validate the combination of parameters passed in.
+
+        // This parser has been passed a null filename and the GATK is not responsible for creating a type default for the object;
+        // therefore, the user must have failed to specify a type default
+        if(writerFileName == null) {
+            if(!source.isRequired())
+                throw new MissingArgumentValueException(bamArgumentDefinition);
+            if(generateMD5)
+                throw new ArgumentException("MD5 generation specified, but no output file specified.  If md5 generation is desired, please specify a BAM output file and an md5 file will be written alongside.");
+        }
+
+        // Create the stub and set parameters.
+        SAMFileWriterStub stub = new SAMFileWriterStub(engine, new File(writerFileName));
+
         if( compressionLevel != null )
             stub.setCompressionLevel(compressionLevel);
-
-        stub.setIndexOnTheFly(!argumentIsPresent(disableWriteIndexArgumentDefinition(source),matches));
-        stub.setSimplifyBAM(argumentIsPresent(createSimplifyBAMArgumentDefinition(source),matches));
+        if(indexOnTheFly != null)
+            stub.setIndexOnTheFly(indexOnTheFly);
+        if(generateMD5 != null)
+            stub.setGenerateMD5(generateMD5);
+        if(simplifyBAM != null)
+            stub.setSimplifyBAM(simplifyBAM);
 
         // WARNING: Side effects required by engine!
         parsingEngine.addTags(stub,getArgumentTags(matches));
@@ -187,6 +205,23 @@ public class SAMFileWriterArgumentTypeDescriptor extends ArgumentTypeDescriptor
                                        null );
     }
 
+    private ArgumentDefinition enableMD5GenerationArgumentDefinition(ArgumentSource source) {
+        return new ArgumentDefinition( ArgumentIOType.ARGUMENT,
+                                       boolean.class,
+                                       ENABLE_MD5_FULLNAME,
+                                       null,
+                                       "Enable on-the-fly creation of md5s for output BAM files.",
+                                       false,
+                                       true,
+                                       false,
+                                       source.isHidden(),
+                                       null,
+                                       null,
+                                       null,
+                                       null );
+    }
+
+
     private ArgumentDefinition createSimplifyBAMArgumentDefinition(ArgumentSource source) {
         return new ArgumentDefinition( ArgumentIOType.ARGUMENT,
                                        boolean.class,
diff --git a/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterStub.java b/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterStub.java
index e11f53a4b..f5c1e0efc 100644
--- a/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterStub.java
+++ b/java/src/org/broadinstitute/sting/gatk/io/stubs/SAMFileWriterStub.java
@@ -77,6 +77,11 @@ public class SAMFileWriterStub implements Stub<SAMFileWriter>, StingSAMFileWrite
      */
     private boolean indexOnTheFly = false;
 
+    /**
+     * Should the GATK generate an md5 for the output BAM?
+     */
+    private boolean generateMD5 = false;
+
     /**
      * Should this BAM be presorted?
      */
@@ -198,6 +203,24 @@ public class SAMFileWriterStub implements Stub<SAMFileWriter>, StingSAMFileWrite
         this.indexOnTheFly = indexOnTheFly;
     }
 
+    /**
+     * Gets whether to generate an md5 on-the-fly for this BAM.
+     * @return True generates the md5.  False means skip writing the file.
+     */
+    public Boolean getGenerateMD5() {
+        return generateMD5;
+    }
+
+    /**
+     * Gets whether to generate an md5 on-the-fly for this BAM.
+     * @return True generates the md5.  False means skip writing the file.
+     */
+    public void setGenerateMD5(boolean generateMD5) {
+        if(writeStarted)
+            throw new UserException("Attempted to turn on md5 generation for BAM file with alignments already in it.");        
+        this.generateMD5 = generateMD5;
+    }
+
     /**
      * Whether the BAM file to create is actually presorted.
      * @return True if the BAM file is presorted.  False otherwise.