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

Feature allow custom adapters

fixes #266 

See merge request !384

flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/AdapterSequence.scala

+package nl.lumc.sasc.biopet.pipelines.flexiprep
+
+/** Case class representing a known adapter sequence */
+case class AdapterSequence(name: String, seq: String)

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

@@ -16,7 +16,6 @@
 package nl.lumc.sasc.biopet.pipelines.flexiprep
 
 import nl.lumc.sasc.biopet.utils.config.Configurable
-import scala.collection.JavaConversions._
 
 /**
  * Cutadapt wrapper specific for Flexiprep.
@@ -30,9 +29,29 @@ import scala.collection.JavaConversions._
  */
 class Cutadapt(root: Configurable, fastqc: Fastqc) extends nl.lumc.sasc.biopet.extensions.Cutadapt(root) {
 
+  val ignoreFastqcAdapters: Boolean = config("ignore_fastqc_adapters", default = false)
+  val customAdaptersConfig: Map[String, Any] = config("custom_adapters", default = Map.empty)
+
   /** Clipped adapter names from FastQC */
-  protected def seqToName: Map[String, String] = fastqc.foundAdapters
-    .map(adapter => adapter.seq -> adapter.name).toMap
+  protected def seqToName: Map[String, String] = {
+    if (!ignoreFastqcAdapters) {
+      (fastqc.foundAdapters ++ customAdapters)
+        .map(adapter => adapter.seq -> adapter.name).toMap
+    } else {
+      customAdapters.map(adapter => adapter.seq -> adapter.name).toMap
+    }
+  }
+
+  def customAdapters: Set[AdapterSequence] = {
+    customAdaptersConfig.flatMap(adapter => {
+      adapter match {
+        case (adapterName: String, sequence: String) =>
+          Some(AdapterSequence(adapterName, sequence))
+        case _ =>
+          throw new IllegalStateException(s"Custom adapter was setup wrong for: $adapter")
+      }
+    }).toSet
+  }
 
   override def summaryStats: Map[String, Any] = {
     val initStats = super.summaryStats

flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Fastqc.scala

@@ -150,9 +150,6 @@ class Fastqc(root: Configurable) extends nl.lumc.sasc.biopet.extensions.Fastqc(r
       }
     } else Map()
 
-  /** Case class representing a known adapter sequence */
-  protected case class AdapterSequence(name: String, seq: String)
-
   /**
    * Retrieves overrepresented sequences found by FastQ.
    *

flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/QcCommand.scala

@@ -102,9 +102,15 @@ class QcCommand(val root: Configurable, val fastqc: Fastqc) extends BiopetComman
     addPipeJob(seqtk)
 
     clip = if (!flexiprep.skipClip) {
-      val foundAdapters = fastqc.foundAdapters.map(_.seq)
+      val cutadapt = clip.getOrElse(new Cutadapt(root, fastqc))
+
+      val foundAdapters = if (!cutadapt.ignoreFastqcAdapters) {
+        fastqc.foundAdapters.map(_.seq) ++ cutadapt.customAdapters.map(_.seq)
+      } else {
+        cutadapt.customAdapters.map(_.seq)
+      }
+
       if (foundAdapters.nonEmpty) {
-        val cutadapt = clip.getOrElse(new Cutadapt(root, fastqc))
         cutadapt.fastqInput = seqtk.output
         cutadapt.fastqOutput = new File(output.getParentFile, input.getName + ".cutadapt.fq")
         cutadapt.statsOutput = new File(flexiprep.outputDir, s"${flexiprep.sampleId.getOrElse("x")}-${flexiprep.libId.getOrElse("x")}.$read.clip.stats")