Merge branch 'develop' into feature-allow-custom-adapters

Conflicts:
	flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Cutadapt.scala

bam2wig/pom.xml

@@ -45,6 +45,18 @@
             <artifactId>BiopetExtensions</artifactId>
             <version>${project.version}</version>
         </dependency>
+        <dependency>
+            <groupId>org.scalatest</groupId>
+            <artifactId>scalatest_2.10</artifactId>
+            <version>2.2.1</version>
+            <scope>test</scope>
+        </dependency>
+        <dependency>
+            <groupId>org.testng</groupId>
+            <artifactId>testng</artifactId>
+            <version>6.8</version>
+            <scope>test</scope>
+        </dependency>
     </dependencies>
 
 </project>
\ No newline at end of file

bam2wig/src/test/scala/nl/lumc/sasc/biopet/pipelines/bamtobigwig/BamToChromSizesTest.scala

+package nl.lumc.sasc.biopet.pipelines.bamtobigwig
+
+import java.io.File
+import java.nio.file.Paths
+
+import org.scalatest.Matchers
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.io.Source
+
+/**
+ * Created by pjvanthof on 09/05/16.
+ */
+class BamToChromSizesTest extends TestNGSuite with Matchers {
+  private def resourcePath(p: String): String = {
+    Paths.get(getClass.getResource(p).toURI).toString
+  }
+
+  @Test
+  def testChromSizes: Unit = {
+    val bamFile = new File(resourcePath("/empty.bam"))
+    val bamToChromSizes = new BamToChromSizes(null)
+    bamToChromSizes.bamFile = bamFile
+    bamToChromSizes.chromSizesFile = File.createTempFile("chrom.", ".sizes")
+    bamToChromSizes.chromSizesFile.deleteOnExit()
+    bamToChromSizes.run()
+    Source.fromFile(bamToChromSizes.chromSizesFile).getLines().toList shouldBe List("chrQ\t10000", "chrR\t10000")
+  }
+}

biopet-core/src/main/scala/nl/lumc/sasc/biopet/core/WriteDependencies.scala

@@ -30,18 +30,16 @@ import scala.collection.mutable.ListBuffer
  */
 object WriteDependencies extends Logging with Configurable {
   val root: Configurable = null
-  private val functionNames: mutable.Map[QFunction, String] = mutable.Map()
-
-  private def createFunctionNames(functions: Seq[QFunction]): Unit = {
+  private def createFunctionNames(functions: Seq[QFunction]): Map[QFunction, String] = {
     val cache: mutable.Map[String, Int] = mutable.Map()
-    for (function <- functions) {
+    (for (function <- functions) yield {
       val baseName = function match {
         case f: Configurable => f.configNamespace
         case f               => f.getClass.getSimpleName
       }
       cache += baseName -> (cache.getOrElse(baseName, 0) + 1)
-      functionNames += function -> s"$baseName-${cache(baseName)}"
-    }
+      function -> s"$baseName-${cache(baseName)}"
+    }).toMap
   }
 
   /**
@@ -55,7 +53,7 @@ object WriteDependencies extends Logging with Configurable {
 
     val errorOnMissingInput: Boolean = config("error_on_missing_input", false)
 
-    createFunctionNames(functions)
+    val functionNames = createFunctionNames(functions)
 
     case class QueueFile(file: File) {
       private val inputJobs: ListBuffer[QFunction] = ListBuffer()

biopet-core/src/test/scala/nl/lumc/sasc/biopet/core/WriteDependenciesTest.scala

+package nl.lumc.sasc.biopet.core
+
+import java.io.File
+import java.nio.file.Files
+
+import nl.lumc.sasc.biopet.utils.ConfigUtils
+import org.broadinstitute.gatk.queue.function.QFunction
+import org.scalatest.Matchers
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.io.Source
+
+/**
+ * Created by pjvanthof on 09/05/16.
+ */
+class WriteDependenciesTest extends TestNGSuite with Matchers {
+
+  import WriteDependenciesTest._
+
+  case class Qfunc(in: Seq[File], out: Seq[File]) extends QFunction {
+    override def inputs = in
+    override def outputs = out
+    override def doneOutputs = out.map(x => new File(x.getParentFile, s".${x.getName}.done"))
+    override def failOutputs = out.map(x => new File(x.getParentFile, s".${x.getName}.fail"))
+    jobOutputFile = new File(out.head + ".out")
+  }
+
+  @Test
+  def testDeps: Unit = {
+    val outputFile = File.createTempFile("deps.", ".json")
+    outputFile.deleteOnExit()
+    val func1 = Qfunc(file1 :: Nil, file2 :: Nil)
+    val func2 = Qfunc(file2 :: Nil, file3 :: Nil)
+    WriteDependencies.writeDependencies(func1 :: func2 :: Nil, outputFile)
+    val deps = ConfigUtils.fileToConfigMap(outputFile)
+    deps("jobs") shouldBe a[Map[_, _]]
+    val jobs = deps("jobs").asInstanceOf[Map[String, Map[String, Any]]]
+    jobs.count(_._1.contains("Qfunc")) shouldBe 2
+
+    deps("files") shouldBe a[List[_]]
+    val files = deps("files").asInstanceOf[List[Map[String, Any]]]
+    val paths = files.map(x => x.get("path")).flatten
+    assert(paths.contains(file1.toString))
+    assert(paths.contains(file2.toString))
+    assert(paths.contains(file3.toString))
+
+    files.find(_.get("path") == Some(file1.toString)).flatMap(_.get("pipeline_input")) shouldBe Some(true)
+    files.find(_.get("path") == Some(file2.toString)).flatMap(_.get("pipeline_input")) shouldBe Some(false)
+    files.find(_.get("path") == Some(file3.toString)).flatMap(_.get("pipeline_input")) shouldBe Some(false)
+  }
+}
+
+object WriteDependenciesTest {
+  val tempDir = Files.createTempDirectory("test").toFile
+  tempDir.deleteOnExit()
+  val file1 = new File(tempDir, "file1.txt")
+  val file2 = new File(tempDir, "file2.txt")
+  val file3 = new File(tempDir, "file3.txt")
+}
\ No newline at end of file

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/Cutadapt.scala

@@ -24,6 +24,7 @@ import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
 
 import scala.collection.mutable
 import scala.io.Source
+import scala.util.matching.Regex
 
 /**
  * Extension for cutadapt
@@ -163,6 +164,51 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
     (if (outputAsStsout) "" else required("--output", fastqOutput) +
       " > " + required(statsOutput))
 
+  def extractClippedAdapters(statsOutput: File): Map[String, Any] = {
+    val histoCountRow: Regex = """([\d]+)\t([\d]+)\t.*""".r
+    val adapterR = """Sequence: ([C|T|A|G]+);.*Trimmed: ([\d]+) times\.""".r
+
+    val statsFile = Source.fromFile(statsOutput)
+    val adapterRawStats: Array[String] = statsFile.mkString
+      .split("=== Adapter [\\d]+ ===")
+      .filter(_.contains("Sequence")
+      )
+    statsFile.close()
+
+    adapterRawStats.map(adapter => {
+      var adapterName = ""
+      var adapterCount = 0
+      // identify the adapter name and count
+      for (line <- adapter.split("\n")) {
+        line match {
+          case adapterR(adapter, count) => {
+            adapterName = adapter
+            adapterCount = count.toInt
+          }
+          case _ =>
+        }
+      }
+
+      // parse the block that gives the histogram of clipped bases and from which end
+      val counts = adapter.split("Overview of removed sequences ")
+        .filter(x => x.contains("length"))
+        .map(clipSideRawStats => {
+          val clipSideLabel = if (clipSideRawStats.contains("5'")) { "5p" } else { "3p" }
+
+          val histogramValues = clipSideRawStats.split("\n").flatMap({
+            case histoCountRow(length, count) => Some(length.toInt -> count.toInt)
+            case _                            => None
+          })
+          clipSideLabel -> histogramValues.toMap
+        })
+
+      adapterName -> Map(
+        "count" -> adapterCount,
+        "histogram" -> counts.toMap
+      )
+    }).toMap // converting the Array[String] containing map-items to Map with 'toMap'
+  }
+
   /** Output summary stats */
   def summaryStats: Map[String, Any] = {
     /**
@@ -177,7 +223,6 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
     val tooLongR = """.* that were too long: *([,\d]+) .*""".r
 
     val tooManyN = """.* with too many N: *([,\d]+) .*""".r
-    val adapterR = """Sequence ([C|T|A|G]*);.*Trimmed: ([,\d]+) times.""".r
 
     val basePairsProcessed = """Total basepairs processed: *([,\d]+) bp""".r
     val basePairsWritten = """Total written \(filtered\): *([,\d]+) bp .*""".r
@@ -192,24 +237,28 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
       "bpoutput" -> 0,
       "toomanyn" -> 0
     )
-    val adapterStats: mutable.Map[String, Long] = mutable.Map()
+
+    // extract the adapters with its histogram
+    val adapterStats = if (statsOutput.exists) {
+      extractClippedAdapters(statsOutput)
+    } else Map.empty
 
     if (statsOutput.exists) {
       val statsFile = Source.fromFile(statsOutput)
       for (line <- statsFile.getLines()) {
         line match {
-          case processedReads(m)        => stats("processed") = m.replaceAll(",", "").toLong
-          case withAdapters(m)          => stats("withadapters") = m.replaceAll(",", "").toLong
-          case readsPassingFilters(m)   => stats("passingfilters") = m.replaceAll(",", "").toLong
-          case tooShortR(m)             => stats("tooshort") = m.replaceAll(",", "").toLong
-          case tooLongR(m)              => stats("toolong") = m.replaceAll(",", "").toLong
-          case tooManyN(m)              => stats("toomanyn") = m.replaceAll(",", "").toLong
-          case basePairsProcessed(m)    => stats("bpinput") = m.replaceAll(",", "").toLong
-          case basePairsWritten(m)      => stats("bpoutput") = m.replaceAll(",", "").toLong
-          case adapterR(adapter, count) => adapterStats += (adapter -> count.toLong)
-          case _                        =>
+          case processedReads(m)      => stats("processed") = m.replaceAll(",", "").toLong
+          case withAdapters(m)        => stats("withadapters") = m.replaceAll(",", "").toLong
+          case readsPassingFilters(m) => stats("passingfilters") = m.replaceAll(",", "").toLong
+          case tooShortR(m)           => stats("tooshort") = m.replaceAll(",", "").toLong
+          case tooLongR(m)            => stats("toolong") = m.replaceAll(",", "").toLong
+          case tooManyN(m)            => stats("toomanyn") = m.replaceAll(",", "").toLong
+          case basePairsProcessed(m)  => stats("bpinput") = m.replaceAll(",", "").toLong
+          case basePairsWritten(m)    => stats("bpoutput") = m.replaceAll(",", "").toLong
+          case _                      =>
         }
       }
+      statsFile.close()
     }
 
     val cleanReads = stats("processed") - stats("withadapters")
@@ -223,8 +272,8 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
       "num_reads_discarded_too_long" -> stats("toolong"),
       "num_reads_discarded_many_n" -> stats("toomanyn"),
       "num_bases_input" -> stats("bpinput"),
-      "num_based_output" -> stats("bpoutput"),
-      adaptersStatsName -> adapterStats.toMap
+      "num_bases_output" -> stats("bpoutput"),
+      adaptersStatsName -> adapterStats
     )
   }
 

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/CombineGVCFs.scala

@@ -69,7 +69,7 @@ class CombineGVCFs(val root: Configurable) extends CommandLineGATK with ScatterG
   override def cmdLine = super.cmdLine +
     repeat("-A", annotation, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-G", group, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse(null)), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-V", variant, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +
     conditional(convertToBasePairResolution, "-bpResolution", escape = true, format = "%s") +

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/GenotypeGVCFs.scala

@@ -106,16 +106,12 @@ class GenotypeGVCFs(val root: Configurable) extends CommandLineGATK with Scatter
   @Gather(enabled = false)
   private var outputIndex: File = _
 
-  @Output
-  @Gather(enabled = false)
-  private var dbsnpIndex: File = _
-
   override def beforeGraph() {
     super.beforeGraph()
     deps ++= variant.filter(orig => orig != null && (!orig.getName.endsWith(".list"))).map(orig => VcfUtils.getVcfIndexFile(orig))
     if (out != null && !org.broadinstitute.gatk.utils.io.IOUtils.isSpecialFile(out))
       outputIndex = VcfUtils.getVcfIndexFile(out)
-    dbsnp.foreach(x => dbsnpIndex = VcfUtils.getVcfIndexFile(x))
+    dbsnp.foreach(x => deps :+= VcfUtils.getVcfIndexFile(x))
   }
 
   override def cmdLine = super.cmdLine +
@@ -133,7 +129,7 @@ class GenotypeGVCFs(val root: Configurable) extends CommandLineGATK with Scatter
     optional("-ploidy", sample_ploidy, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-A", annotation, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-G", group, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse(null)), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
     conditional(filter_reads_with_N_cigar, "-filterRNC", escape = true, format = "%s") +
     conditional(filter_mismatching_base_and_quals, "-filterMBQ", escape = true, format = "%s") +
     conditional(filter_bases_not_stored, "-filterNoBases", escape = true, format = "%s")

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/HaplotypeCaller.scala

@@ -419,7 +419,7 @@ class HaplotypeCaller(val root: Configurable) extends CommandLineGATK with Scatt
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +
     optional("-likelihoodEngine", likelihoodCalculationEngine, spaceSeparated = true, escape = true, format = "%s") +
     optional("-hksr", heterogeneousKmerSizeResolution, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse(null)), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
     conditional(dontTrimActiveRegions, "-dontTrimActiveRegions", escape = true, format = "%s") +
     optional("-maxDiscARExtension", maxDiscARExtension, spaceSeparated = true, escape = true, format = "%s") +
     optional("-maxGGAARExtension", maxGGAARExtension, spaceSeparated = true, escape = true, format = "%s") +
@@ -444,7 +444,7 @@ class HaplotypeCaller(val root: Configurable) extends CommandLineGATK with Scatt
     repeat("-inputPrior", input_prior, spaceSeparated = true, escape = true, format = "%s") +
     optional("-ploidy", sample_ploidy, spaceSeparated = true, escape = true, format = "%s") +
     optional("-gt_mode", genotyping_mode, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-alleles", alleles), alleles, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-alleles", alleles.getOrElse(null)), alleles, spaceSeparated = true, escape = true, format = "%s") +
     optional("-contamination", contamination_fraction_to_filter, spaceSeparated = true, escape = true, format = contamination_fraction_to_filterFormat) +
     optional("-contaminationFile", contamination_fraction_per_sample_file, spaceSeparated = true, escape = true, format = "%s") +
     optional("-pnrm", p_nonref_model, spaceSeparated = true, escape = true, format = "%s") +

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/SelectVariants.scala

@@ -218,8 +218,8 @@ class SelectVariants(val root: Configurable) extends CommandLineGATK with Scatte
 
   override def cmdLine = super.cmdLine +
     required(TaggedFile.formatCommandLineParameter("-V", variant), variant, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-disc", discordance), discordance, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-conc", concordance), concordance, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-disc", discordance.getOrElse(null)), discordance, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-conc", concordance.getOrElse(null)), concordance, spaceSeparated = true, escape = true, format = "%s") +
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-sn", sample_name, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-se", sample_expressions, spaceSeparated = true, escape = true, format = "%s") +

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/UnifiedGenotyper.scala

@@ -276,7 +276,7 @@ class UnifiedGenotyper(val root: Configurable) extends CommandLineGATK with Scat
     conditional(ignoreSNPAlleles, "-ignoreSNPAlleles", escape = true, format = "%s") +
     conditional(allReadsSP, "-dl", escape = true, format = "%s") +
     conditional(ignoreLaneInfo, "-ignoreLane", escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-referenceCalls", reference_sample_calls), reference_sample_calls, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-referenceCalls", reference_sample_calls.getOrElse(null)), reference_sample_calls, spaceSeparated = true, escape = true, format = "%s") +
     optional("-refsample", reference_sample_name, spaceSeparated = true, escape = true, format = "%s") +
     optional("-minqs", min_quality_score, spaceSeparated = true, escape = true, format = "%s") +
     optional("-maxqs", max_quality_score, spaceSeparated = true, escape = true, format = "%s") +
@@ -291,14 +291,14 @@ class UnifiedGenotyper(val root: Configurable) extends CommandLineGATK with Scat
     repeat("-inputPrior", input_prior, spaceSeparated = true, escape = true, format = "%s") +
     optional("-ploidy", sample_ploidy, spaceSeparated = true, escape = true, format = "%s") +
     optional("-gt_mode", genotyping_mode, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-alleles", alleles), alleles, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-alleles", alleles.getOrElse(null)), alleles, spaceSeparated = true, escape = true, format = "%s") +
     optional("-contamination", contamination_fraction_to_filter, spaceSeparated = true, escape = true, format = contamination_fraction_to_filterFormat) +
     optional("-contaminationFile", contamination_fraction_per_sample_file, spaceSeparated = true, escape = true, format = "%s") +
     optional("-pnrm", p_nonref_model, spaceSeparated = true, escape = true, format = "%s") +
     optional("-logExactCalls", exactcallslog, spaceSeparated = true, escape = true, format = "%s") +
     optional("-out_mode", output_mode, spaceSeparated = true, escape = true, format = "%s") +
     conditional(allSitePLs, "-allSitePLs", escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse(null)), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-comp", comp, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-onlyEmitSamples", onlyEmitSamples, spaceSeparated = true, escape = true, format = "%s") +

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/VariantAnnotator.scala

@@ -104,8 +104,8 @@ class VariantAnnotator(val root: Configurable) extends CommandLineGATK with Scat
 
   override def cmdLine = super.cmdLine +
     required(TaggedFile.formatCommandLineParameter("-V", variant), variant, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-snpEffFile", snpEffFile), snpEffFile, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-snpEffFile", snpEffFile.getOrElse()), snpEffFile, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse()), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-comp", comp, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-resource", resource, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +

biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/gatk/VariantEval.scala

@@ -135,8 +135,8 @@ class VariantEval(val root: Configurable) extends CommandLineGATK {
     optional("-o", out, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-eval", eval, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-comp", comp, formatPrefix = TaggedFile.formatCommandLineParameter, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
-    optional(TaggedFile.formatCommandLineParameter("-gold", goldStandard), goldStandard, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-D", dbsnp.getOrElse(null)), dbsnp, spaceSeparated = true, escape = true, format = "%s") +
+    optional(TaggedFile.formatCommandLineParameter("-gold", goldStandard.getOrElse(null)), goldStandard, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-select", select_exps, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-selectName", select_names, spaceSeparated = true, escape = true, format = "%s") +
     repeat("-sn", sample, spaceSeparated = true, escape = true, format = "%s") +

biopet-tools-extensions/src/test/scala/nl/lumc/sasc/biopet/extensions/tools/VcfFilterTest.scala

@@ -19,7 +19,7 @@ import java.io.File
 
 import org.scalatest.Matchers
 import org.scalatest.testng.TestNGSuite
-import org.testng.annotations.{ DataProvider, Test }
+import org.testng.annotations.Test
 
 /**
  * Created by ahbbollen on 2-3-16.
@@ -44,34 +44,26 @@ class VcfFilterTest extends TestNGSuite with Matchers {
     filterer.outputVcfIndex.getAbsolutePath shouldBe oVcf.getAbsolutePath + ".tbi"
   }
 
-  @DataProvider(name = "functions")
-  def functions = {
-    Array(
-      () => testCommand(minSampleDepth = Some(2)),
-      () => testCommand(minTotalDepth = Some(2)),
-      () => testCommand(minAlternateDepth = Some(2)),
-      () => testCommand(minSamplesPass = Some(2)),
-      () => testCommand(minGenomeQuality = Some(50)),
-      () => testCommand(filterRefCalls = true),
-      () => testCommand(invertedOutputVcf = Some(File.createTempFile("vcfFilter", ".vcf"))),
-      () => testCommand(resToDom = Some("dummy")),
-      () => testCommand(trioCompound = Some("dummy")),
-      () => testCommand(deNovoInSample = Some("dummy")),
-      () => testCommand(deNovoTrio = Some("dummy")),
-      () => testCommand(trioLossOfHet = Some("dummy")),
-      () => testCommand(mustHaveVariant = List("sample1", "sample2")),
-      () => testCommand(calledIn = List("sample1", "sample2")),
-      () => testCommand(mustHaveGenotype = List("sample1:HET", "sample2:HET")),
-      () => testCommand(diffGenotype = List("sample1:sample2", "sample2:sample3")),
-      () => testCommand(minQualScore = Some(50.0)),
-      () => testCommand(filterHetVarToHomVar = List("dummy")),
-      () => testCommand(id = List("rs01", "rs02")),
-      () => testCommand(idFile = Some(File.createTempFile("vcfFilter", ".txt")))
-    ).map(Array(_))
-  }
-
-  @Test(dataProvider = "functions")
-  def executer(function0: Function0[Unit]): Unit = function0()
+  @Test def testMinSampleDepth() = testCommand(minSampleDepth = Some(2))
+  @Test def testMinTotalDepth() = testCommand(minTotalDepth = Some(2))
+  @Test def testMinAlternateDepth() = testCommand(minAlternateDepth = Some(2))
+  @Test def testMinSamplesPass() = testCommand(minSamplesPass = Some(2))
+  @Test def testMinGenomeQuality() = testCommand(minGenomeQuality = Some(50))
+  @Test def testFilterRefCalls() = testCommand(filterRefCalls = true)
+  @Test def testInvertedOutputVcf() = testCommand(invertedOutputVcf = Some(File.createTempFile("vcfFilter", ".vcf")))
+  @Test def testResToDom() = testCommand(resToDom = Some("dummy"))
+  @Test def testTrioCompound() = testCommand(trioCompound = Some("dummy"))
+  @Test def testDeNovoInSample() = testCommand(deNovoInSample = Some("dummy"))
+  @Test def testDeNovoTrio() = testCommand(deNovoTrio = Some("dummy"))
+  @Test def testTrioLossOfHet() = testCommand(trioLossOfHet = Some("dummy"))
+  @Test def testMustHaveVariant() = testCommand(mustHaveVariant = List("sample1", "sample2"))
+  @Test def testCalledIn() = testCommand(calledIn = List("sample1", "sample2"))
+  @Test def testMustHaveGenotype() = testCommand(mustHaveGenotype = List("sample1:HET", "sample2:HET"))
+  @Test def testDiffGenotype() = testCommand(diffGenotype = List("sample1:sample2", "sample2:sample3"))
+  @Test def testMinQualScore() = testCommand(minQualScore = Some(50.0))
+  @Test def testFilterHetVarToHomVar() = testCommand(filterHetVarToHomVar = List("dummy"))
+  @Test def testId() = testCommand(id = List("rs01", "rs02"))
+  @Test def testIdFile() = testCommand(idFile = Some(File.createTempFile("vcfFilter", ".txt")))
 
   protected def testCommand(minSampleDepth: Option[Int] = None,
                             minTotalDepth: Option[Int] = None,

biopet-utils/src/main/scala/nl/lumc/sasc/biopet/utils/summary/Summary.scala

@@ -61,7 +61,7 @@ class Summary(file: File) {
   }
 
   /** Executes given function for each sample */
-  def getSampleValues(function: (Summary, String) => Option[Any]): Map[String, Option[Any]] = {
+  def getSampleValues[T](function: (Summary, String) => Option[T]): Map[String, Option[T]] = {
     (for (sample <- samples) yield sample -> function(this, sample)).toMap
   }
 
@@ -95,7 +95,7 @@ class Summary(file: File) {
    * @param function Function to execute
    * @return (sampleId, libId) -> value
    */
-  def getLibraryValues(function: (Summary, String, String) => Option[Any]): Map[(String, String), Option[Any]] = {
+  def getLibraryValues[T](function: (Summary, String, String) => Option[T]): Map[(String, String), Option[T]] = {
     (for (sample <- samples; lib <- libraries.getOrElse(sample, Set())) yield {
       (sample, lib) -> function(this, sample, lib)
     }).toMap

biopet-utils/src/main/scala/nl/lumc/sasc/biopet/utils/summary/SummaryValue.scala

@@ -33,7 +33,6 @@ case class SummaryValue(value: Option[Any]) {
     })
   }
 
-  //TODO: Calculations are not yet used somewhere, needs more testing
   def +(that: SummaryValue): SummaryValue = {
     (this.value, that.value) match {
       case (Some(a: Double), Some(b)) => SummaryValue(Some(a + b.toString.toDouble))

biopet-utils/src/test/scala/nl/lumc/sasc/biopet/utils/IoUtilsTest.scala

+package nl.lumc.sasc.biopet.utils
+
+import java.io.{ File, FileNotFoundException, PrintWriter }
+import java.nio.file.Files
+
+import org.scalatest.Matchers
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.io.Source
+
+/**
+ * Created by pjvanthof on 05/05/16.
+ */
+class IoUtilsTest extends TestNGSuite with Matchers {
+
+  def createTempTestFile(file: File): Unit = {
+    file.getParentFile.mkdirs()
+    val writer = new PrintWriter(file)
+    writer.println("test")
+    writer.close()
+    file.deleteOnExit()
+  }
+
+  @Test
+  def testCopyFile: Unit = {
+    val temp1 = File.createTempFile("test.", ".txt")
+    temp1.deleteOnExit()
+    val temp2 = File.createTempFile("test.", ".txt")
+    temp2.deleteOnExit()
+    createTempTestFile(temp1)
+    IoUtils.copyFile(temp1, temp2)
+    val reader = Source.fromFile(temp2)
+    reader.getLines().toList shouldBe List("test")
+    reader.close()
+  }
+
+  @Test
+  def testCopyFileNonExistingDir: Unit = {
+    val temp1 = File.createTempFile("test.", ".txt")
+    val tempDir = new File(Files.createTempDirectory("test").toFile, "non-exist")
+    tempDir.deleteOnExit()
+    tempDir shouldNot exist
+    val temp2 = new File(tempDir, "test.txt")
+    createTempTestFile(temp1)
+    intercept[FileNotFoundException] {
+      IoUtils.copyFile(temp1, temp2)
+    }
+    IoUtils.copyFile(temp1, temp2, true)
+    val reader = Source.fromFile(temp2)
+    reader.getLines().toList shouldBe List("test")
+    reader.close()
+  }
+
+  @Test
+  def testCopyDir: Unit = {
+    val tempDir1 = Files.createTempDirectory("test").toFile