In the new sharding system, if no read group is present, hallucinate one. Added

for test compatibility, but not sure whether we still need this feature.  TODO: Poll the group about this feature.


git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@2949 348d0f76-0448-11de-a6fe-93d51630548a

build.xml

@@ -305,8 +305,8 @@
             <jvmarg value="-javaagent:${build.dir}/TestAgent.jar"/>
             <formatter type="${testOutput}" usefile="${usefile}" />
             <classpath>
-                <path refid="runtime.dependencies"/>
                 <pathelement location="build"/>
+                <path refid="runtime.dependencies"/>
                 <pathelement location="${test.classes}"/>
                 <pathelement location="lib/junit-4.4.jar"/>
             </classpath>

java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/BlockDrivenSAMDataSource.java

@@ -415,6 +415,20 @@ public class BlockDrivenSAMDataSource extends SAMDataSource {
             for(File readsFile: sourceInfo.getReadsFiles()) {
                 SAMFileReader2 reader = new SAMFileReader2(readsFile,true);
                 reader.setValidationStringency(sourceInfo.getValidationStringency());
+
+                // If no read group is present, hallucinate one.
+                // TODO: Straw poll to see whether this is really required.
+                final SAMFileHeader header = reader.getFileHeader();
+                logger.debug(String.format("Sort order is: " + header.getSortOrder()));
+
+                if (reader.getFileHeader().getReadGroups().size() < 1) {
+                    SAMReadGroupRecord rec = new SAMReadGroupRecord(readsFile.getName());
+                    rec.setLibrary(readsFile.getName());
+                    rec.setSample(readsFile.getName());
+
+                    reader.getFileHeader().addReadGroup(rec);
+                }
+
                 readers.put(new SAMReaderID(readsFile),reader);
             }
         }

java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/IndexDrivenSAMDataSource.java

@@ -415,7 +415,7 @@ public class IndexDrivenSAMDataSource extends SAMDataSource {
     private StingSAMIterator createIterator( DataStreamSegment segment ) {
         StingSAMIterator iterator = resourcePool.iterator(segment);
         StingSAMIterator malformedWrappedIterator =  new MalformedSAMFilteringIterator( getHeader(), iterator, violations );
-        StingSAMIterator readWrappingIterator = new ReadWrappingIterator(malformedWrappedIterator);
+        StingSAMIterator readWrappingIterator = new ReadFormattingIterator(malformedWrappedIterator);
         return readWrappingIterator;
     }
 }

java/src/org/broadinstitute/sting/gatk/datasources/simpleDataSources/SAMDataSource.java

@@ -12,7 +12,6 @@ import org.broadinstitute.sting.utils.sam.SAMReadViolationHistogram;
 
 import java.io.File;
 import java.util.Collection;
-import java.util.Map;
 import java.util.List;
 import java.util.ArrayList;
 
@@ -182,7 +181,7 @@ public abstract class SAMDataSource implements SimpleDataSource {
                                                         Boolean noValidationOfReadOrder,
                                                         Collection<SamRecordFilter> supplementalFilters) {
         wrappedIterator = new MalformedSAMFilteringIterator(getHeader(),wrappedIterator,violations );
-        wrappedIterator = new ReadWrappingIterator(wrappedIterator);
+        wrappedIterator = new ReadFormattingIterator(wrappedIterator);
 
         // NOTE: this (and other filtering) should be done before on-the-fly sorting
         //  as there is no reason to sort something that we will end of throwing away

java/src/org/broadinstitute/sting/gatk/iterators/ReadFormattingIterator.java

@@ -1,17 +1,27 @@
 package org.broadinstitute.sting.gatk.iterators;
 
 import net.sf.samtools.SAMRecord;
+import net.sf.samtools.SAMTag;
+import net.sf.samtools.SAMReadGroupRecord;
 import org.broadinstitute.sting.gatk.Reads;
 import org.broadinstitute.sting.utils.sam.GATKSAMRecord;
+import org.apache.log4j.Logger;
+
+import java.util.List;
 
 /**
- * An iterator which wraps each SAMRecord inside a wrapper class, bringing new functionality to the read while
- * presenting the original SAMRecord interface.
+ * An iterator which does post-processing of a read, including potentially wrapping
+ * the read in something with a compatible interface or replacing the read entirely.
  *
  * @author mhanna
  * @version 0.1
  */
-public class ReadWrappingIterator implements StingSAMIterator {
+public class ReadFormattingIterator implements StingSAMIterator {
+    /**
+     * Logger.
+     */
+    protected static Logger logger = Logger.getLogger(ReadFormattingIterator.class);    
+
     /**
      * Iterator to which to pass
      */
@@ -21,7 +31,7 @@ public class ReadWrappingIterator implements StingSAMIterator {
      * Decorate the given iterator inside a ReadWrappingIterator.
      * @param wrappedIterator iterator
      */
-    public ReadWrappingIterator(StingSAMIterator wrappedIterator) {
+    public ReadFormattingIterator(StingSAMIterator wrappedIterator) {
         this.wrappedIterator = wrappedIterator;
     }
 
@@ -65,7 +75,21 @@ public class ReadWrappingIterator implements StingSAMIterator {
      *         no next exists.
      */
     public SAMRecord next() {
-        return new GATKSAMRecord(wrappedIterator.next());
+        SAMRecord read = wrappedIterator.next();
+
+        // if we don't have a read group, set one.
+        // TODO: Straw poll to see whether this is really required.        
+        if (read.getAttribute(SAMTag.RG.toString()) == null && read.getReader() != null) {
+            List<SAMReadGroupRecord> readGroups = read.getReader().getFileHeader().getReadGroups();
+            if (readGroups.size() == 1) {
+                read.setAttribute(SAMTag.RG.toString(), readGroups.get(0).getReadGroupId());
+                read.setAttribute(SAMTag.SM.toString(), readGroups.get(0).getReadGroupId());
+            } else {
+                logger.warn("Unable to set read group of ungrouped read: unable to pick default group, there are " + readGroups.size() + " possible.");
+            }
+        }
+
+        return new GATKSAMRecord(read);
     }
 
     /**