Seqstat now uses quality type data

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

@@ -75,24 +75,24 @@ class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
     outputFiles += ("fastq_input_R1" -> extractIfNeeded(input_R1, outputDir))
     if (paired) outputFiles += ("fastq_input_R2" -> extractIfNeeded(input_R2, outputDir))
 
-    addSeqstat(outputFiles("fastq_input_R1"), "seqstat_R1")
-    if (paired) addSeqstat(outputFiles("fastq_input_R2"), "seqstat_R2")
-
-    addSha1sum(outputFiles("fastq_input_R1"), "sha1_R1")
-    if (paired) addSha1sum(outputFiles("fastq_input_R2"), "sha1_R2")
-
     var fastqc_R1 = Fastqc(this, input_R1, outputDir + "/" + R1_name + ".fastqc/")
     add(fastqc_R1)
     outputFiles += ("fastqc_R1" -> fastqc_R1.output)
     outputFiles += ("qualtype_R1" -> getQualtype(fastqc_R1, R1_name))
     outputFiles += ("contams_R1" -> getContams(fastqc_R1, R1_name))
-
+    
+    addSeqstat(outputFiles("fastq_input_R1"), "seqstat_R1", fastqc_R1)
+    addSha1sum(outputFiles("fastq_input_R1"), "sha1_R1")
+    
     if (paired) {
       var fastqc_R2 = Fastqc(this, input_R2, outputDir + "/" + R2_name + ".fastqc/")
       add(fastqc_R2)
       outputFiles += ("fastqc_R2" -> fastqc_R2.output)
       outputFiles += ("qualtype_R2" -> getQualtype(fastqc_R2, R2_name))
       outputFiles += ("contams_R2" -> getContams(fastqc_R2, R2_name))
+      
+      addSeqstat(outputFiles("fastq_input_R2"), "seqstat_R2", fastqc_R2)
+      addSha1sum(outputFiles("fastq_input_R2"), "sha1_R2")
     }
   }
 
@@ -275,11 +275,13 @@ class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
     } else return file
   }
 
-  def addSeqstat(fastq: File, key: String) {
+  def addSeqstat(fastq: File, key: String, fastqc:Fastqc = null) {
     val ext = fastq.getName.substring(fastq.getName.lastIndexOf("."))
     val seqstat = new Seqstat(this)
     seqstat.input_fastq = fastq
+    seqstat.fastqc = fastqc
     seqstat.out = swapExt(outputDir, fastq, ext, ".seqstats.json")
+    if (fastqc != null) seqstat.deps ::= fastqc.output
     add(seqstat)
     outputFiles += (key -> seqstat.out)
   }

flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/scripts/Seqstat.scala

@@ -2,6 +2,7 @@ package nl.lumc.sasc.biopet.pipelines.flexiprep.scripts
 
 import java.io.File
 
+import nl.lumc.sasc.biopet.function.fastq.Fastqc
 import org.broadinstitute.gatk.utils.commandline.{ Input, Output }
 
 import nl.lumc.sasc.biopet.core.config.Configurable
@@ -16,9 +17,23 @@ class Seqstat(val root: Configurable) extends PythonCommandLineFunction {
 
   @Output(doc = "Output file", shortName = "out", required = true)
   var out: File = _
-
+  
   var fmt: String = _
-
+  var fastqc: Fastqc = _
+  
+  override def beforeCmd {
+    if (fastqc != null && fmt == null) {
+      fastqc.getEncoding match {
+        case null => {}
+        case s if (s.contains("Sanger / Illumina 1.9")) => fmt = "sanger"
+        case s if (s.contains("Illumina <1.3")) => fmt = "solexa"
+        case s if (s.contains("Illumina 1.3")) => fmt = "illumina"
+        case s if (s.contains("Illumina 1.5")) => fmt = "illumina"
+        //case _ => null
+      }
+    }
+  }
+  
   def cmdLine = {
     getPythonCommand +
       optional("--fmt", fmt) +