Merge branch 'feature-update-cutadapt' into 'develop'

Update Cutadapt wrapper

Updated cutadapt wrapper to reflect cutadapt 1.9 and added missing options (which were already supported in 1.5 but not in extension)

Closes #243 

See merge request !310

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

@@ -26,18 +26,19 @@ import scala.collection.mutable
 import scala.io.Source
 
 /**
- * Extension for cutadept
- * Based on version 1.5
+ * Extension for cutadapt
+ * Started with version 1.5
+ * Updated to version 1.9 (18-01-2016 by wyleung)
  */
 class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Summarizable with Version {
   @Input(doc = "Input fastq file")
-  var fastq_input: File = _
+  var fastqInput: File = _
 
   @Output
-  var fastq_output: File = _
+  var fastqOutput: File = _
 
   @Output(doc = "Output statistics file")
-  var stats_output: File = _
+  var statsOutput: File = _
 
   executable = config("exe", default = "cutadapt")
   def versionCommand = executable + " --version"
@@ -46,28 +47,121 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
   /** Name of the key containing clipped adapters information in the summary stats. */
   def adaptersStatsName = "adapters"
 
-  var default_clip_mode: String = config("default_clip_mode", default = "3")
-  var opt_adapter: Set[String] = config("adapter", default = Nil)
-  var opt_anywhere: Set[String] = config("anywhere", default = Nil)
-  var opt_front: Set[String] = config("front", default = Nil)
-
-  var opt_discard: Boolean = config("discard", default = false)
-  var opt_minimum_length: Int = config("minimum_length", 1)
-  var opt_maximum_length: Option[Int] = config("maximum_length")
+  var defaultClipMode: String = config("default_clip_mode", default = "3")
+  var adapter: Set[String] = config("adapter", default = Nil)
+  var anywhere: Set[String] = config("anywhere", default = Nil)
+  var front: Set[String] = config("front", default = Nil)
+
+  var errorRate: Option[Double] = config("error_rate")
+  var noIndels: Boolean = config("no_indels", default = false)
+  var times: Option[Int] = config("times")
+  var overlap: Option[Int] = config("overlap")
+  var matchReadWildcards: Boolean = config("match_read_wildcards", default = false)
+  var noMatchAdapterWildcards: Boolean = config("no_match_adapter_wildcards", default = false) // specific for 1.9
+
+  /** Options for filtering of processed reads */
+  var discard: Boolean = config("discard", default = false)
+  var trimmedOnly: Boolean = config("trimmed_only", default = false)
+  var minimumLength: Int = config("minimum_length", 1)
+  var maximumLength: Option[Int] = config("maximum_length")
+  var noTrim: Boolean = config("no_trim", default = false)
+  var maxN: Option[Int] = config("max_n") // specific for 1.9
+  var maskAdapter: Boolean = config("mask_adapter", default = false)
+
+  /** Options that influence what gets output to where */
+  var quiet: Boolean = config("quiet", default = false)
+  //  var output: File // see up @Output
+  var infoFile: Option[File] = config("info_file")
+  var restFile: Option[File] = config("rest_file")
+  var wildcardFile: Option[File] = config("wildcard_file")
+  var tooShortOutput: Option[File] = config("too_short_output")
+  var tooLongOutput: Option[File] = config("too_long_output")
+  var untrimmedOutput: Option[File] = config("untrimmed_output")
+
+  /** Additional read modifications */
+  var cut: Option[Int] = config("cut")
+  var qualityCutoff: Option[String] = config("quality_cutoff")
+  var qualityBase: Option[Int] = config("quality_base")
+  var trimN: Boolean = config("trim_n", default = false)
+  var prefix: Option[String] = config("prefix")
+  var suffix: Option[String] = config("suffix")
+  var stripSuffix: Set[String] = config("strip_suffix")
+  var lengthTag: Option[String] = config("length_tag")
+
+  /** Colorspace options */
+  var colorspace: Boolean = config("colorspace", default = false)
+  var doubleEncode: Boolean = config("double_encode", default = false)
+  var trimPrimer: Boolean = config("trim_primer", default = false)
+  var stripF3: Boolean = config("strip_f3", default = false)
+  var maq: Boolean = config("maq", default = false)
+  var bwa: Boolean = config("bwa", default = false)
+  var noZeroCap: Boolean = config("no_zero_cap", default = false)
+  var zeroCap: Boolean = config("zero_cap", default = false)
+
+  /** Paired end options */
+  var peAdapter: Set[String] = config("pe_adapter", default = Nil)
+  var peAdapterFront: Set[String] = config("pe_adapter_front", default = Nil)
+  var peAdapterBoth: Set[String] = config("pe_adapter_both", default = Nil)
+  var peCut: Boolean = config("pe_cut", default = false)
+  var pairedOutput: Option[File] = config("paired_output")
+  var interleaved: Boolean = config("interleaved", default = false)
+  var untrimmedPairedOutput: Option[File] = config("untrimmed_paired_output")
 
   /** return commandline to execute */
   def cmdLine = required(executable) +
-    // options
-    repeat("-a", opt_adapter) +
-    repeat("-b", opt_anywhere) +
-    repeat("-g", opt_front) +
-    conditional(opt_discard, "--discard") +
-    optional("-m", opt_minimum_length) +
-    optional("-M", opt_maximum_length) +
+    // Options that influence how the adapters are found
+    repeat("-a", adapter) +
+    repeat("-b", anywhere) +
+    repeat("-g", front) +
+    optional("--error-rate", errorRate) +
+    conditional(noIndels, "--no-indels") +
+    optional("--times", times) +
+    optional("--overlap", overlap) +
+    conditional(matchReadWildcards, "--match-read-wildcards") +
+    conditional(noMatchAdapterWildcards, "--no-match-adapter-wildcards") +
+    // Options for filtering of processed reads
+    conditional(discard, "--discard") +
+    conditional(trimmedOnly, "--trimmed-only") +
+    optional("-m", minimumLength) +
+    optional("-M", maximumLength) +
+    conditional(noTrim, "--no-trim") +
+    optional("--max-n", maxN) +
+    conditional(maskAdapter, "--mask-adapter") +
+    conditional(quiet, "--quiet") +
+    optional("--info-file", infoFile) +
+    optional("--rest-file", restFile) +
+    optional("--wildcard-file", wildcardFile) +
+    optional("--too-short-output", tooShortOutput) +
+    optional("--too-long-output", tooLongOutput) +
+    optional("--untrimmed-output", untrimmedOutput) +
+    // Additional read modifications
+    optional("--cut", cut) +
+    optional("--quality-cutoff", qualityCutoff) +
+    conditional(trimN, "--trim-n") +
+    optional("--prefix", prefix) +
+    optional("--suffix", suffix) +
+    optional("--strip-suffix", stripSuffix) +
+    optional("--length-tag", lengthTag) +
+    // Colorspace options
+    conditional(colorspace, "--colorspace") +
+    conditional(doubleEncode, "--double-encode") +
+    conditional(trimPrimer, "--trim-primer") +
+    conditional(stripF3, "--strip-f3") +
+    conditional(maq, "--maq") +
+    conditional(bwa, "--bwa") +
+    conditional(noZeroCap, "--no-zero-cap") +
+    conditional(zeroCap, "--zero-cap") +
+    // Paired-end options
+    repeat("-A", peAdapter) +
+    repeat("-G", peAdapterFront) +
+    repeat("-B", peAdapterBoth) +
+    conditional(interleaved, "--interleaved") +
+    optional("--paired-output", pairedOutput) +
+    optional("--untrimmed-paired-output", untrimmedPairedOutput) +
     // input / output
-    required(fastq_input) +
-    (if (outputAsStsout) "" else required("--output", fastq_output) +
-      " > " + required(stats_output))
+    required(fastqInput) +
+    (if (outputAsStsout) "" else required("--output", fastqOutput) +
+      " > " + required(statsOutput))
 
   /** Output summary stats */
   def summaryStats: Map[String, Any] = {
@@ -79,7 +173,7 @@ class Cutadapt(val root: Configurable) extends BiopetCommandLineFunction with Su
     val stats: mutable.Map[String, Int] = mutable.Map("trimmed" -> 0, "tooshort" -> 0, "toolong" -> 0)
     val adapter_stats: mutable.Map[String, Int] = mutable.Map()
 
-    if (stats_output.exists) for (line <- Source.fromFile(stats_output).getLines()) {
+    if (statsOutput.exists) for (line <- Source.fromFile(statsOutput).getLines()) {
       line match {
         case trimR(m)                 => stats += ("trimmed" -> m.toInt)
         case tooShortR(m)             => stats += ("tooshort" -> m.toInt)

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

@@ -46,16 +46,16 @@ class Cutadapt(root: Configurable, fastqc: Fastqc) extends nl.lumc.sasc.biopet.e
             // adapter sequence is clipped but not found by FastQC ~ should not happen since all clipped adapter
             // sequences come from FastQC
             case _ =>
-              throw new IllegalStateException(s"Adapter '$seq' is clipped but not found by FastQC in '$fastq_input'.")
+              throw new IllegalStateException(s"Adapter '$seq' is clipped but not found by FastQC in '$fastqInput'.")
           }
         // FastQC found no adapters
         case otherwise =>
           ;
-          logger.debug(s"No adapters found for summarizing in '$fastq_input'.")
+          logger.debug(s"No adapters found for summarizing in '$fastqInput'.")
           None
       }
       // "adapters" key not found ~ something went wrong in our part
-      case _ => throw new RuntimeException(s"Required key 'adapters' not found in stats entry '$fastq_input'.")
+      case _ => throw new RuntimeException(s"Required key 'adapters' not found in stats entry '$fastqInput'.")
     }
     initStats.updated(adaptersStatsName, adapterCounts)
   }

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

@@ -102,12 +102,12 @@ class QcCommand(val root: Configurable, val fastqc: Fastqc) extends BiopetComman
       val foundAdapters = fastqc.foundAdapters.map(_.seq)
       if (foundAdapters.nonEmpty) {
         val cutadapt = new Cutadapt(root, fastqc)
-        cutadapt.fastq_input = seqtk.output
-        cutadapt.fastq_output = new File(output.getParentFile, input.getName + ".cutadapt.fq")
-        cutadapt.stats_output = new File(flexiprep.outputDir, s"${flexiprep.sampleId.getOrElse("x")}-${flexiprep.libId.getOrElse("x")}.$read.clip.stats")
-        if (cutadapt.default_clip_mode == "3") cutadapt.opt_adapter ++= foundAdapters
-        else if (cutadapt.default_clip_mode == "5") cutadapt.opt_front ++= foundAdapters
-        else if (cutadapt.default_clip_mode == "both") cutadapt.opt_anywhere ++= foundAdapters
+        cutadapt.fastqInput = seqtk.output
+        cutadapt.fastqOutput = new File(output.getParentFile, input.getName + ".cutadapt.fq")
+        cutadapt.statsOutput = new File(flexiprep.outputDir, s"${flexiprep.sampleId.getOrElse("x")}-${flexiprep.libId.getOrElse("x")}.$read.clip.stats")
+        if (cutadapt.defaultClipMode == "3") cutadapt.adapter ++= foundAdapters
+        else if (cutadapt.defaultClipMode == "5") cutadapt.front ++= foundAdapters
+        else if (cutadapt.defaultClipMode == "both") cutadapt.anywhere ++= foundAdapters
         addPipeJob(cutadapt)
         Some(cutadapt)
       } else None
@@ -117,7 +117,7 @@ class QcCommand(val root: Configurable, val fastqc: Fastqc) extends BiopetComman
       val sickle = new Sickle(root)
       sickle.output_stats = new File(flexiprep.outputDir, s"${flexiprep.sampleId.getOrElse("x")}-${flexiprep.libId.getOrElse("x")}.$read.trim.stats")
       sickle.input_R1 = clip match {
-        case Some(c) => c.fastq_output
+        case Some(c) => c.fastqOutput
         case _       => seqtk.output
       }
       sickle.output_R1 = new File(output.getParentFile, input.getName + ".sickle.fq")
@@ -127,7 +127,7 @@ class QcCommand(val root: Configurable, val fastqc: Fastqc) extends BiopetComman
 
     val outputFile = (clip, trim) match {
       case (_, Some(t)) => t.output_R1
-      case (Some(c), _) => c.fastq_output
+      case (Some(c), _) => c.fastqOutput
       case _            => seqtk.output
     }