Merge 69dcfa5 into 511f925

bigdatagenomics · Aug 30, 2017 · ae21c0b · ae21c0b
2 parents 511f925 + 69dcfa5
commit ae21c0b
Show file tree

Hide file tree

Showing 3 changed files with 27 additions and 5 deletions.
diff --git a/adam-core/src/main/scala/org/bdgenomics/adam/rdd/fragment/FragmentRDD.scala b/adam-core/src/main/scala/org/bdgenomics/adam/rdd/fragment/FragmentRDD.scala
@@ -93,6 +93,14 @@ object FragmentRDD {
    */
   val WRITE_SUFFIXES = "org.bdgenomics.adam.rdd.fragment.FragmentRDD.writeSuffixes"
 
+  /**
+   * Hadoop configuration path to check for a boolean value indicating that
+   * the warning logging that is triggered if an interleaved fragment doesn't
+   * have exactly two reads should be disabled. Default is false (logging is
+   * enabled).
+   */
+  val DISABLE_READ_COUNT_LOGGING = "org.bdgenomics.adam.rdd.fragment.FragmentRDD.disableReadCountLogging"
+
   /**
    * Creates a FragmentRDD where no record groups or sequence info are attached.
    *

diff --git a/adam-core/src/main/scala/org/bdgenomics/adam/rdd/fragment/InterleavedFASTQInFormatter.scala b/adam-core/src/main/scala/org/bdgenomics/adam/rdd/fragment/InterleavedFASTQInFormatter.scala
@@ -48,6 +48,7 @@ class InterleavedFASTQInFormatter private (
   private val converter = new AlignmentRecordConverter
   private val writeSuffixes = conf.getBoolean(FragmentRDD.WRITE_SUFFIXES, false)
   private val writeOriginalQualities = conf.getBoolean(FragmentRDD.WRITE_ORIGINAL_QUALITIES, false)
+  private val enableLogging = !conf.getBoolean(FragmentRDD.DISABLE_READ_COUNT_LOGGING, false)
 
   /**
    * Writes alignment records to an output stream in interleaved FASTQ format.
@@ -59,11 +60,11 @@ class InterleavedFASTQInFormatter private (
     iter.flatMap(frag => {
       val reads = converter.convertFragment(frag).toSeq
 
-      if (reads.size < 2) {
+      if (enableLogging && reads.size < 2) {
         log.warn("Fewer than two reads for %s. Dropping...".format(frag))
         None
       } else {
-        if (reads.size > 2) {
+        if (enableLogging && reads.size > 2) {
           log.warn("More than two reads for %s. Taking first 2.".format(frag))
         }
         Some((reads(0), reads(1)))
@@ -80,8 +81,20 @@ class InterleavedFASTQInFormatter private (
         outputOriginalBaseQualities = writeOriginalQualities) + "\n"
 
       // write both to the output stream
-      os.write(fastq1.getBytes)
-      os.write(fastq2.getBytes)
+      // ensure that reads are ordered properly if ordering is known (see #1702)
+      if (read1.getReadInFragment == 0 &&
+        read2.getReadInFragment == 1) {
+        os.write(fastq1.getBytes)
+        os.write(fastq2.getBytes)
+      } else if (read1.getReadInFragment == 1 &&
+        read2.getReadInFragment == 0) {
+        os.write(fastq2.getBytes)
+        os.write(fastq1.getBytes)
+      } else {
+        if (enableLogging) {
+          log.warn("Improper pair of reads in fragment %s. Dropping...".format(p))
+        }
+      }
     })
   }
 }
diff --git a/adam-core/src/main/scala/org/bdgenomics/adam/rdd/read/AnySAMOutFormatter.scala b/adam-core/src/main/scala/org/bdgenomics/adam/rdd/read/AnySAMOutFormatter.scala
@@ -29,7 +29,7 @@ import scala.collection.mutable.ListBuffer
  * An OutFormatter that automatically infers whether the piped input is SAM or
  * BAM. Autodetecting streamed CRAM is not currently supported.
  */
-class AnySAMOutFormatter extends OutFormatter[AlignmentRecord] {
+case class AnySAMOutFormatter(stringency: ValidationStringency = ValidationStringency.STRICT) extends OutFormatter[AlignmentRecord] {
 
   /**
    * Reads alignment records from an input stream. Autodetects SAM/BAM format.
@@ -41,6 +41,7 @@ class AnySAMOutFormatter extends OutFormatter[AlignmentRecord] {
 
     // make reader
     val reader = SamReaderFactory.makeDefault()
+      .validationStringency(stringency)
       .open(SamInputResource.of(is))
 
     SAMIteratorConverter(reader)