Biopet core code inspection round

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

@@ -44,7 +44,7 @@ trait SummaryQScript extends BiopetQScript { qscript: QScript =>
   private[summary] var summaryQScripts: List[SummaryQScript] = Nil
 
   /** Name of the pipeline in the summary */
-  var summaryName = configNamespace
+  var summaryName: String = configNamespace
 
   /** Must return a map with used settings for this pipeline */
   def summarySettings: Map[String, Any]
@@ -98,7 +98,7 @@ trait SummaryQScript extends BiopetQScript { qscript: QScript =>
       Await.result(
         SummaryDb.openSqliteSummary(summaryDbFile).getRuns(runId = Some(id)).map(_.headOption),
         Duration.Inf) match {
-        case Some(x) => id
+        case Some(_) => id
         case _ =>
           logger.warn(
             s"Run id found in '$runIdFile' does not exist in summary, creating a new run")
@@ -191,22 +191,22 @@ trait SummaryQScript extends BiopetQScript { qscript: QScript =>
     qscript match {
       case q: MultiSampleQScript =>
         // Global level
-        for ((key, file) <- qscript.summaryFiles) addChecksum(file)
+        for ((_, file) <- qscript.summaryFiles) addChecksum(file)
 
-        for ((sampleName, sample) <- q.samples) {
+        for ((_, sample) <- q.samples) {
           // Sample level
-          for ((key, file) <- sample.summaryFiles) addChecksum(file)
-          for ((libName, lib) <- sample.libraries) {
+          for ((_, file) <- sample.summaryFiles) addChecksum(file)
+          for ((_, lib) <- sample.libraries) {
             // Library level
-            for ((key, file) <- lib.summaryFiles) addChecksum(file)
+            for ((_, file) <- lib.summaryFiles) addChecksum(file)
           }
         }
-      case q => for ((key, file) <- q.summaryFiles) addChecksum(file)
+      case q => for ((_, file) <- q.summaryFiles) addChecksum(file)
     }
 
     for (inputFile <- inputFiles) {
       inputFile.md5 match {
-        case Some(checksum) => {
+        case Some(checksum) =>
           val checkMd5 = new CheckChecksum
           checkMd5.inputFile = inputFile.file
           if (!SummaryQScript.md5sumCache.contains(inputFile.file))
@@ -216,7 +216,6 @@ trait SummaryQScript extends BiopetQScript { qscript: QScript =>
           checkMd5.jobOutputFile = new File(checkMd5.checksumFile.getParentFile,
                                             checkMd5.checksumFile.getName + ".check.out")
           add(checkMd5)
-        }
         case _ =>
       }
     }

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

@@ -25,7 +25,7 @@ import org.broadinstitute.gatk.utils.commandline.Input
 
 import scala.collection.mutable
 import scala.io.Source
-import scala.concurrent.Await
+import scala.concurrent.{Await, Future}
 import scala.concurrent.duration.Duration
 import scala.concurrent.ExecutionContext.Implicits.global
 
@@ -40,7 +40,7 @@ class WriteSummary(val parent: SummaryQScript) extends InProcessFunction with Co
   require(parent != null)
 
   /** To access qscript for this summary */
-  val qscript = parent
+  val qscript: SummaryQScript = parent
 
   @Input(doc = "deps", required = false)
   var deps: List[File] = Nil
@@ -62,28 +62,27 @@ class WriteSummary(val parent: SummaryQScript) extends InProcessFunction with Co
     val db = SummaryDb.openSqliteSummary(qscript.summaryDbFile)
     if (qscript == root) { // This initialize the database
       qscript match {
-        case s: MultiSampleQScript => s.initSummaryDb
+        case s: MultiSampleQScript => s.initSummaryDb()
         case t: SampleLibraryTag =>
           t.sampleId.foreach {
-            case sampleName =>
+            sampleName =>
               val sampleId = Await
                 .result(db.getSamples(name = Some(sampleName), runId = Some(qscript.summaryRunId))
-                          .map(_.headOption.map(_.id)),
-                        Duration.Inf)
+                  .map(_.headOption.map(_.id)),
+                  Duration.Inf)
                 .getOrElse {
                   Await.result(db.createOrUpdateSample(sampleName, qscript.summaryRunId),
-                               Duration.Inf)
+                    Duration.Inf)
                 }
               t.libId.foreach { libName =>
-                val libId = Await
-                  .result(db.getSamples(name = Some(libName),
-                                        runId = Some(qscript.summaryRunId),
-                                        sampleId = Some(sampleId))
-                            .map(_.headOption.map(_.id)),
-                          Duration.Inf)
+                Await.result(db.getSamples(name = Some(libName),
+                  runId = Some(qscript.summaryRunId),
+                  sampleId = Some(sampleId))
+                  .map(_.headOption.map(_.id)),
+                  Duration.Inf)
                   .getOrElse {
                     Await.result(db.createOrUpdateLibrary(libName, qscript.summaryRunId, sampleId),
-                                 Duration.Inf)
+                      Duration.Inf)
                   }
               }
           }
@@ -102,7 +101,7 @@ class WriteSummary(val parent: SummaryQScript) extends InProcessFunction with Co
           try {
             deps :+= f.firstOutput
           } catch {
-            case e: NullPointerException => logger.debug("Queue values are not initialized")
+            case _: NullPointerException => logger.debug("Queue values are not initialized")
           }
         case _ =>
       }
@@ -328,7 +327,7 @@ class WriteSummary(val parent: SummaryQScript) extends InProcessFunction with Co
   }
 
   /** Convert summarizable to a summary map */
-  def parseSummarizable(summarizable: Summarizable, name: String) = {
+  def parseSummarizable(summarizable: Summarizable, name: String): Map[String, Any] = {
     val stats = summarizable.summaryStats
     val files = parseFiles(summarizable.summaryFiles)
 
@@ -365,11 +364,11 @@ object WriteSummary {
                  libId: Option[Int],
                  key: String,
                  file: File,
-                 outputDir: File) = {
+                 outputDir: File): Future[Int] = {
     val path = if (file.getAbsolutePath.startsWith(outputDir.getAbsolutePath + File.separator)) {
       "." + file.getAbsolutePath.stripPrefix(s"${outputDir.getAbsolutePath}")
     } else file.getAbsolutePath
-    val md5 = SummaryQScript.md5sumCache.get(file).map(WriteSummary.parseChecksum(_))
+    val md5 = SummaryQScript.md5sumCache.get(file).map(WriteSummary.parseChecksum)
     val size = if (file.exists()) file.length() else 0L
     val link = if (file.exists()) java.nio.file.Files.isSymbolicLink(file.toPath) else false
     db.createOrUpdateFile(runId,

biopet-core/src/main/scala/nl/lumc/sasc/biopet/core/workaround/BiopetQCommandLine.scala

@@ -98,7 +98,7 @@ object BiopetQCommandLine extends GatkLogging {
       System.exit(CommandLineProgram.result)
   }
 
-  val timestamp = System.currentTimeMillis
+  val timestamp: Long = System.currentTimeMillis
 }
 
 /**
@@ -135,7 +135,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
 
     // override createByType to pass the correct exceptions
     new PluginManager[QScript](qPluginType, List(qScriptClasses.toURI.toURL)) {
-      override def createByType(plugintype: Class[_ <: QScript]) = {
+      override def createByType(plugintype: Class[_ <: QScript]): QScript = {
         val noArgsConstructor = plugintype.getDeclaredConstructor()
         noArgsConstructor.setAccessible(true)
         noArgsConstructor.newInstance()
@@ -157,7 +157,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
     * Takes the QScripts passed in, runs their script() methods, retrieves their generated
     * functions, and then builds and runs a QGraph based on the dependencies.
     */
-  def execute = {
+  def execute: Int = {
     var success = false
     var result = 1
     var functionsAndStatusSize = 0
@@ -181,7 +181,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
       }
 
       qGraph.messengers =
-        allCommandPlugins.filter(_.statusMessenger != null).map(_.statusMessenger).toSeq
+        allCommandPlugins.filter(_.statusMessenger != null).map(_.statusMessenger)
 
       // TODO: Default command plugin argument?
       val remoteFileConverter =
@@ -270,7 +270,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
     * so that their arguments can be inspected before QScript.script is called.
     * @return Array of dynamic sources
     */
-  override def getArgumentSources = {
+  override def getArgumentSources: Array[Class[_]] = {
     var plugins = Seq.empty[Class[_]]
     plugins ++= qScriptPluginManager.getPlugins
     plugins ++= qCommandPlugin.getPlugins
@@ -281,7 +281,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
     * Returns the name of a script/plugin
     * @return The name of a script/plugin
     */
-  override def getArgumentSourceName(source: Class[_]) = {
+  override def getArgumentSourceName(source: Class[_]): String = {
     if (classOf[QScript].isAssignableFrom(source))
       qScriptPluginManager.getName(source.asSubclass(classOf[QScript]))
     else if (classOf[QCommandPlugin].isAssignableFrom(source))
@@ -294,7 +294,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
     * Returns a ScalaCompoundArgumentTypeDescriptor that can parse argument sources into scala collections.
     * @return a ScalaCompoundArgumentTypeDescriptor
     */
-  override def getArgumentTypeDescriptors =
+  override def getArgumentTypeDescriptors: util.List[ArgumentTypeDescriptor] =
     util.Arrays.asList(new ScalaCompoundArgumentTypeDescriptor)
 
   override def getApplicationDetails: ApplicationDetails = {
@@ -338,7 +338,7 @@ class BiopetQCommandLine extends CommandLineProgram with Logging {
     }
   }
 
-  def shutdown() = {
+  def shutdown(): AnyVal = {
     shuttingDown = true
     qGraph.shutdown()
     if (qScriptClasses != null) IOUtils.tryDelete(qScriptClasses)

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/BamStats.scala

@@ -39,8 +39,8 @@ class BamStats(val parent: Configurable)
 
   var outputDir: File = _
 
-  var binSize: Option[Int] = config("bin_size")
-  var threadBinSize: Option[Int] = config("thread_bin_size")
+  val binSize: Option[Int] = config("bin_size")
+  val threadBinSize: Option[Int] = config("thread_bin_size")
 
   override def defaultThreads = 3
   override def defaultCoreMemory = 5.0
@@ -68,7 +68,7 @@ class BamStats(val parent: Configurable)
   }
 
   /** Creates command to execute extension */
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("-b", bamFile) +
       required("-o", outputDir) +

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/BastyGenerateFasta.scala

@@ -43,8 +43,8 @@ class BastyGenerateFasta(val parent: Configurable) extends ToolCommandFunction w
 
   var snpsOnly: Boolean = config("snps_only", default = false)
   var sampleName: String = _
-  var minAD: Int = config("min_ad", default = 8)
-  var minDepth: Int = config("min_depth", default = 8)
+  val minAD: Int = config("min_ad", default = 8)
+  val minDepth: Int = config("min_depth", default = 8)
   var outputName: String = _
 
   override def defaultCoreMemory = 4.0
@@ -54,7 +54,7 @@ class BastyGenerateFasta(val parent: Configurable) extends ToolCommandFunction w
     reference = referenceFasta()
   }
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       optional("--inputVcf", inputVcf) +
       optional("--bamFile", bamFile) +

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/BedToInterval.scala

@@ -37,7 +37,7 @@ class BedToInterval(val parent: Configurable) extends ToolCommandFunction {
 
   override def defaultCoreMemory = 1.0
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine + required("-I", input) + required("-b", bamFile) + required("-o", output)
 }
 

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/BedtoolsCoverageToCounts.scala

@@ -31,7 +31,7 @@ class BedtoolsCoverageToCounts(val parent: Configurable) extends ToolCommandFunc
 
   override def defaultCoreMemory = 3.0
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("-I", input) +
       required("-o", output)

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/BiopetFlagstat.scala

@@ -37,7 +37,7 @@ class BiopetFlagstat(val parent: Configurable) extends ToolCommandFunction with
 
   override def defaultCoreMemory = 6.0
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine + required("-I", input) + required("-s", summaryFile) + " > " + required(output)
 
   def summaryFiles: Map[String, File] = Map()

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/CheckValidateAnnotation.scala

@@ -40,7 +40,7 @@ class CheckValidateAnnotation(val parent: Configurable)
   var genomeName: String = ""
 
   /** Exits whenever the input md5sum is not the same as the output md5sum */
-  def run: Unit = {
+  def run(): Unit = {
 
     val reader = Source.fromFile(inputLogFile)
     reader.getLines().foreach { line =>

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/DownloadNcbiAssembly.scala

@@ -29,7 +29,7 @@ class DownloadNcbiAssembly(val parent: Configurable) extends ToolCommandFunction
   var outputReport: File = _
 
   @Input(required = true)
-  var assemblyReport: File = null
+  var assemblyReport: File = _
 
   var nameHeader: Option[String] = None
 
@@ -38,7 +38,7 @@ class DownloadNcbiAssembly(val parent: Configurable) extends ToolCommandFunction
 
   override def defaultCoreMemory = 4.0
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("-a", assemblyReport) +
       required("--report", outputReport) +

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/FastqSplitter.scala

@@ -36,7 +36,7 @@ class FastqSplitter(val parent: Configurable) extends ToolCommandFunction {
   override def defaultCoreMemory = 4.0
 
   /** * Generate command to execute */
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("-I", input) +
       repeat("-o", output)

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/FastqSync.scala

@@ -35,31 +35,31 @@ class FastqSync(val parent: Configurable) extends ToolCommandFunction with Summa
 
   /** Original FASTQ file (read 1 or 2) */
   @Input(required = true)
-  var refFastq: File = null
+  var refFastq: File = _
 
   /** "Input read 1 FASTQ file" */
   @Input(required = true)
-  var inputFastq1: File = null
+  var inputFastq1: File = _
 
   /** Input read 2 FASTQ file */
   @Input(required = true)
-  var inputFastq2: File = null
+  var inputFastq2: File = _
 
   /** Output read 1 FASTQ file */
   @Output(required = true)
-  var outputFastq1: File = null
+  var outputFastq1: File = _
 
   /** Output read 2 FASTQ file */
   @Output(required = true)
-  var outputFastq2: File = null
+  var outputFastq2: File = _
 
   /** Sync statistics */
   @Output(required = true)
-  var outputStats: File = null
+  var outputStats: File = _
 
   override def defaultCoreMemory = 4.0
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("-r", refFastq) +
       required("-i", inputFastq1) +
@@ -98,7 +98,7 @@ class FastqSync(val parent: Configurable) extends ToolCommandFunction with Summa
         "num_reads_kept" -> countRLeft)
   }
 
-  override def summaryDeps = outputStats :: super.summaryDeps
+  override def summaryDeps: List[File] = outputStats :: super.summaryDeps
 
   override def resolveSummaryConflict(v1: Any, v2: Any, key: String): Any = {
     (v1, v2) match {

biopet-core/src/main/scala/nl/lumc/sasc/biopet/extensions/tools/GensToVcf.scala

@@ -63,7 +63,7 @@ class GensToVcf(val parent: Configurable) extends ToolCommandFunction with Refer
     sortInput = true
   }
 
-  override def cmdLine =
+  override def cmdLine: String =
     super.cmdLine +
       required("--inputGenotypes", inputGens) +
       required("--inputInfo", inputInfo) +