diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/Summarizable.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/Summarizable.scala
new file mode 100644
index 0000000000000000000000000000000000000000..f9f36c7e09366c6c04bf84bc625cd86056363e5e
--- /dev/null
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/Summarizable.scala
@@ -0,0 +1,26 @@
+package nl.lumc.sasc.biopet.core.summary
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+
+/**
+ * Created by pjvan_thof on 2/14/15.
+ */
+trait Summarizable extends Configurable {
+
+ var summaryModule = configName
+
+ def summaryFiles: Map[String, File]
+
+ def summaryStats: Map[String, Any]
+
+ /**
+ * This function is used to merge
+ * @param v1
+ * @param v2
+ * @param key
+ * @return
+ */
+ def resolveSummaryConflict(v1: Any, v2: Any, key: String) = v1
+}
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/SummaryQScript.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/SummaryQScript.scala
new file mode 100644
index 0000000000000000000000000000000000000000..acf2727f5f9c7db5512b51061f040e5f60c3d8b2
--- /dev/null
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/SummaryQScript.scala
@@ -0,0 +1,48 @@
+package nl.lumc.sasc.biopet.core.summary
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.BiopetQScript
+import nl.lumc.sasc.biopet.extensions.Md5sum
+
+/**
+ * Created by pjvan_thof on 2/14/15.
+ */
+trait SummaryQScript extends BiopetQScript {
+
+ /** Key is sample/library, None is sample or library is not applicable */
+ private[summary] var summarizables: Map[(Option[String], Option[String]), List[Summarizable]] = Map()
+ private[summary] var summaryQScripts: List[SummaryQScript] = Nil
+
+ def summaryFile: File
+
+ def addSummarizable(summarizable: Summarizable,
+ sampleId: Option[String] = None,
+ libraryId: Option[String] = None): Unit = {
+ require(libraryId.isDefined == sampleId.isDefined) // Library always require a sample
+ summarizables += (sampleId, libraryId) -> (summarizable :: summarizables.getOrElse((sampleId, libraryId), Nil))
+ }
+
+ def addSummaryQScript(summaryQScript: SummaryQScript): Unit = {
+ summaryQScripts :+= summaryQScript
+ }
+
+ def addSummaryJobs: Unit = {
+ val writeSummary = new WriteSummary(this)
+
+ //Automatic checksums
+ val keepChecksums: Boolean = config("keep_checksums_files", default = false)
+
+ for ((_, summarizableList) <- summarizables; summarizable <- summarizableList; (_, file) <- summarizable.summaryFiles) {
+ if (writeSummary.md5sum) {
+ val md5sum = Md5sum(this, file)
+ md5sum.isIntermediate = !keepChecksums
+ writeSummary.deps :+= md5sum.output
+ add(md5sum)
+ }
+ //TODO: add more checksums types
+ }
+
+ add(writeSummary)
+ }
+}
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/WriteSummary.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/WriteSummary.scala
new file mode 100644
index 0000000000000000000000000000000000000000..c573136488a4eecd121ad616b3a2616f18ffe5f3
--- /dev/null
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/core/summary/WriteSummary.scala
@@ -0,0 +1,42 @@
+package nl.lumc.sasc.biopet.core.summary
+
+import java.io.{PrintWriter, File}
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.queue.function.{ QFunction, InProcessFunction }
+import org.broadinstitute.gatk.utils.commandline.{ Output, Input }
+
+/**
+ * Created by pjvan_thof on 2/14/15.
+ */
+class WriteSummary(val root: Configurable) extends InProcessFunction with Configurable {
+ this.analysisName = getClass.getSimpleName
+
+ require(root.isInstanceOf[SummaryQScript], "root is not a SummaryQScript")
+
+ val summaryQScript = root.asInstanceOf[SummaryQScript]
+
+ @Input(doc = "deps", required = false)
+ var deps: List[File] = Nil
+
+ @Output(doc = "Summary output", required = true)
+ var out: File = summaryQScript.summaryFile
+
+ var md5sum: Boolean = config("summary_md5", default = true)
+ //TODO: add more checksums types
+
+ override def freezeFieldValues(): Unit = {
+ for (q <- summaryQScript.summaryQScripts) deps :+= q.summaryFile
+ for ((_, l) <- summaryQScript.summarizables; s <- l) s match {
+ case f: QFunction => deps :+= f.firstOutput
+ case _ =>
+ }
+ super.freezeFieldValues()
+ }
+
+ def run(): Unit = {
+ val writer = new PrintWriter(out)
+
+ writer.close()
+ }
+}
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Md5sum.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Md5sum.scala
index ea2d704f09ce2388c441338ff25b402606288366..091d2b6c39136bd9fcb2c9c36e82f9f4c0301644 100644
--- a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Md5sum.scala
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/Md5sum.scala
@@ -49,4 +49,11 @@ object Md5sum {
md5sum.output = new File(outDir + fastqfile.getName + ".md5")
return md5sum
}
+
+ def apply(root: Configurable, file: File): Md5sum = {
+ val md5sum = new Md5sum(root)
+ md5sum.input = file
+ md5sum.output = new File(file.getParentFile, file.getName + ".md5")
+ return md5sum
+ }
}
diff --git a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Fastqc.scala b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Fastqc.scala
index 21cb069d658623ab5a7b8986e454db6000f927fa..31fdc6b84005d40a40e39ccba110c9e5400899e0 100644
--- a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Fastqc.scala
+++ b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Fastqc.scala
@@ -18,6 +18,8 @@ package nl.lumc.sasc.biopet.pipelines.flexiprep
import java.io.{ File, FileNotFoundException }
+import nl.lumc.sasc.biopet.core.summary.Summarizable
+
import scala.io.Source
import argonaut._, Argonaut._
@@ -32,7 +34,7 @@ import nl.lumc.sasc.biopet.utils.ConfigUtils
* This wrapper implements additional methods for parsing FastQC output files and aggregating everything in a summary
* object. The current implementation is based on FastQC v0.10.1.
*/
-class Fastqc(root: Configurable) extends nl.lumc.sasc.biopet.extensions.Fastqc(root) {
+class Fastqc(root: Configurable) extends nl.lumc.sasc.biopet.extensions.Fastqc(root) with Summarizable {
/** Class for storing a single FastQC module result */
protected case class FastQCModule(name: String, status: String, lines: Seq[String])
@@ -160,6 +162,10 @@ class Fastqc(root: Configurable) extends nl.lumc.sasc.biopet.extensions.Fastqc(r
ConfigUtils.mapToJson(outputMap)
}
+
+ def summaryFiles: Map[String, File] = Map("test" -> this.fastqfile)
+
+ def summaryStats: Map[String, Any] = Map()
}
object Fastqc {
diff --git a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
index 9ab16032bd095d40728e91f92a611366b400af2f..531ed8bbb219da0327795eb295dab842531abf85 100644
--- a/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
+++ b/public/flexiprep/src/main/scala/nl/lumc/sasc/biopet/pipelines/flexiprep/Flexiprep.scala
@@ -15,6 +15,7 @@
*/
package nl.lumc.sasc.biopet.pipelines.flexiprep
+import nl.lumc.sasc.biopet.core.summary.SummaryQScript
import org.broadinstitute.gatk.queue.QScript
import org.broadinstitute.gatk.utils.commandline.{ Input, Argument }
@@ -23,7 +24,7 @@ import nl.lumc.sasc.biopet.core.config.Configurable
import nl.lumc.sasc.biopet.extensions.{ Gzip, Pbzip2, Md5sum, Zcat, Seqstat }
import nl.lumc.sasc.biopet.tools.FastqSync
-class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
+class Flexiprep(val root: Configurable) extends QScript with BiopetQScript with SummaryQScript {
def this() = this(null)
@Input(doc = "R1 fastq file (gzipped allowed)", shortName = "R1", required = true)
@@ -48,6 +49,8 @@ class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
@Argument(doc = "Library ID", shortName = "library", required = true)
var libId: String = _
+ def summaryFile = new File(outputDir, sampleId + "-" + libId + ".qc.summary.json")
+
var paired: Boolean = input_R2.isDefined
var R1_ext: String = _
var R2_ext: String = _
@@ -264,6 +267,7 @@ class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
summary.addMd5sum(md5sum_R2, R2 = true, after = true)
}
fastqc_R1_after = Fastqc(this, R1, outputDir + "/" + R1_name + ".qc.fastqc/")
+ addSummarizable(fastqc_R1_after)
add(fastqc_R1_after)
summary.addFastqc(fastqc_R1_after, after = true)
if (paired) {
@@ -273,7 +277,8 @@ class Flexiprep(val root: Configurable) extends QScript with BiopetQScript {
}
}
- add(summary)
+ //add(summary)
+ addSummaryJobs
}
def extractIfNeeded(file: File, runDir: String): File = {