Adding md5sum checking

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/BiopetQScript.scala

@@ -89,7 +89,7 @@ trait BiopetQScript extends Configurable with GatkLogging {
       globalConfig.writeReport(qSettings.runName, new File(outputDir, ".log/" + qSettings.runName))
     else BiopetQScript.addError("Parent of output dir: '" + outputDir.getParent + "' is not writeable, outputdir can not be created")
 
-    inputFiles.foreach{ i =>
+    inputFiles.foreach { i =>
       if (!i.file.exists()) BiopetQScript.addError(s"Input file does not exist: ${i.file}")
       if (!i.file.canRead()) BiopetQScript.addError(s"Input file can not be read: ${i.file}")
     }

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/SummaryQScript.scala

@@ -18,7 +18,7 @@ package nl.lumc.sasc.biopet.core.summary
 import java.io.File
 
 import nl.lumc.sasc.biopet.core._
-import nl.lumc.sasc.biopet.extensions.Md5sum
+import nl.lumc.sasc.biopet.extensions.{ CheckMd5, Md5sum }
 
 import scala.collection.mutable
 
@@ -27,7 +27,7 @@ import scala.collection.mutable
  *
  * Created by pjvan_thof on 2/14/15.
  */
-trait SummaryQScript extends BiopetQScript {
+trait SummaryQScript extends BiopetQScript { qscript =>
 
   /** Key is sample/library, None is sample or library is not applicable */
   private[summary] var summarizables: Map[(String, Option[String], Option[String]), List[Summarizable]] = Map()
@@ -116,6 +116,20 @@ trait SummaryQScript extends BiopetQScript {
       //TODO: add more checksums types
     }
 
+    for (inputFile <- inputFiles) {
+      inputFile.md5 match {
+        case Some(checksum) => {
+          val checkMd5 = new CheckMd5
+          checkMd5.inputFile = inputFile.file
+          require(SummaryQScript.md5sumCache.contains(inputFile.file), "Md5 job is not executed, checksum file can't be found")
+          checkMd5.md5file = SummaryQScript.md5sumCache(inputFile.file)
+          checkMd5.md5sum = checksum
+          add(checkMd5)
+        }
+        case _ =>
+      }
+    }
+
     for ((_, summarizableList) <- summarizables; summarizable <- summarizableList) {
       summarizable match {
         case f: BiopetCommandLineFunctionTrait => f.beforeGraph()

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/WriteSummary.scala

@@ -153,10 +153,11 @@ class WriteSummary(val root: Configurable) extends InProcessFunction with Config
   def parseFile(file: File): Map[String, Any] = {
     val map: mutable.Map[String, Any] = mutable.Map()
     map += "path" -> file.getAbsolutePath
-    if (md5sum) map += "md5" -> parseChecksum(SummaryQScript.md5sumCache(file))
+    if (md5sum) map += "md5" -> WriteSummary.parseChecksum(SummaryQScript.md5sumCache(file))
     map.toMap
   }
-
+}
+object WriteSummary {
   /** Retrive checksum from file */
   def parseChecksum(checksumFile: File): String = {
     Source.fromFile(checksumFile).getLines().toList.head.split(" ")(0)

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/CheckMd5.scala

+package nl.lumc.sasc.biopet.extensions
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.summary.WriteSummary
+import org.broadinstitute.gatk.queue.function.InProcessFunction
+import org.broadinstitute.gatk.utils.commandline.{ Argument, Input }
+
+/**
+ * This class checks md5sums and give an exit code 1 when md5sum is not the same
+ *
+ * Created by pjvanthof on 16/08/15.
+ */
+class CheckMd5 extends InProcessFunction {
+  @Input(required = true)
+  var inputFile: File = _
+
+  @Input(required = true)
+  var md5file: File = _
+
+  @Argument(required = true)
+  var md5sum: String = _
+
+  /** Exits whenever the input md5sum is not the same as the output md5sum */
+  def run: Unit = {
+    val outputMd5sum = WriteSummary.parseChecksum(md5file).toLowerCase
+
+    if (outputMd5sum != md5sum.toLowerCase) {
+      logger.error(s"Input file: '$inputFile' md5sum is not as expected, aborting pipeline")
+      System.exit(1)
+    }
+  }
+}
\ No newline at end of file