From 9332161d6448ae0ae319d2cea2bf5a338000aa8c Mon Sep 17 00:00:00 2001
From: Peter van 't Hof <p.j.van_t_hof@lumc.nl>
Date: Thu, 16 Apr 2015 13:19:18 +0200
Subject: [PATCH] Switch to multi metrics
---
.../pipelines/bammetrics/BamMetrics.scala | 15 ++--
.../picard/CollectMultipleMetrics.scala | 73 +++++++++++++++++++
2 files changed, 79 insertions(+), 9 deletions(-)
create mode 100644 public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/picard/CollectMultipleMetrics.scala
diff --git a/public/bammetrics/src/main/scala/nl/lumc/sasc/biopet/pipelines/bammetrics/BamMetrics.scala b/public/bammetrics/src/main/scala/nl/lumc/sasc/biopet/pipelines/bammetrics/BamMetrics.scala
index 574adb382..8c94ac09c 100644
--- a/public/bammetrics/src/main/scala/nl/lumc/sasc/biopet/pipelines/bammetrics/BamMetrics.scala
+++ b/public/bammetrics/src/main/scala/nl/lumc/sasc/biopet/pipelines/bammetrics/BamMetrics.scala
@@ -63,15 +63,12 @@ class BamMetrics(val root: Configurable) extends QScript with SummaryQScript wit
add(biopetFlagstat)
addSummarizable(biopetFlagstat, "biopet_flagstat")
- add(CollectGcBiasMetrics(this, inputBam, outputDir))
-
- val collectInsertSizeMetrics = CollectInsertSizeMetrics(this, inputBam, outputDir)
- add(collectInsertSizeMetrics)
- addSummarizable(collectInsertSizeMetrics, "insert_size_metrics")
+ val multiMetrics = new CollectMultipleMetrics(this)
+ multiMetrics.input = inputBam
+ multiMetrics.outputName = new File(outputDir, inputBam.getName.stripSuffix(".bam"))
+ add(multiMetrics)
- val collectAlignmentSummaryMetrics = CollectAlignmentSummaryMetrics(this, inputBam, outputDir)
- add(collectAlignmentSummaryMetrics)
- addSummarizable(collectAlignmentSummaryMetrics, "alignment_metrics")
+ add(CollectGcBiasMetrics(this, inputBam, outputDir))
case class Intervals(bed: File, intervals: File)
@@ -135,7 +132,7 @@ class BamMetrics(val root: Configurable) extends QScript with SummaryQScript wit
}
object BamMetrics extends PipelineCommand {
- /** Make default implementation of BamMetrics */
+ /** Make default implementation of BamMetrics and runs script already */
def apply(root: Configurable, bamFile: File, outputDir: File): BamMetrics = {
val bamMetrics = new BamMetrics(root)
bamMetrics.inputBam = bamFile
diff --git a/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/picard/CollectMultipleMetrics.scala b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/picard/CollectMultipleMetrics.scala
new file mode 100644
index 000000000..e2e7cd49b
--- /dev/null
+++ b/public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/picard/CollectMultipleMetrics.scala
@@ -0,0 +1,73 @@
+package nl.lumc.sasc.biopet.extensions.picard
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.BiopetQScript
+import nl.lumc.sasc.biopet.core.config.Configurable
+import org.broadinstitute.gatk.utils.commandline.{ Output, Argument, Input }
+
+/**
+ * Created by pjvan_thof on 4/16/15.
+ */
+class CollectMultipleMetrics(val root: Configurable) extends Picard {
+ import CollectMultipleMetrics._
+
+ javaMainClass = new picard.analysis.CollectMultipleMetrics().getClass.getName
+
+ @Input(doc = "The input SAM or BAM files to analyze", required = true)
+ var input: File = null
+
+ @Output(doc = "Base name of output files", required = true)
+ var outputName: File = null
+
+ @Argument(doc = "Base name of output files", required = true)
+ var program: List[String] = config("metrics_programs",
+ default = Programs.values.iterator.toList.map(_.toString))
+
+ @Argument(doc = "Assume alignment file is sorted by position", required = false)
+ var assumeSorted: Boolean = config("assume_sorted", default = false)
+
+ @Argument(doc = "Stop after processing N reads", required = false)
+ var stopAfter: Option[Long] = config("stop_after")
+
+ @Output
+ protected var outputFiles: List[File] = Nil
+
+ override def beforeGraph: Unit = {
+ program.foreach(p => p match {
+ case _ if p == Programs.CollectAlignmentSummaryMetrics.toString => {
+ outputFiles :+= new File(outputName + ".alignment_summary_metrics")
+ }
+ case _ if p == Programs.CollectInsertSizeMetrics.toString => {
+ outputFiles :+= new File(outputName + ".insert_size_metrics")
+ outputFiles :+= new File(outputName + ".insert_size_Histogram.pdf")
+ }
+ case _ if p == Programs.QualityScoreDistribution.toString => {
+ outputFiles :+= new File(outputName + ".quality_distribution_metrics")
+ outputFiles :+= new File(outputName + ".test.quality_distribution.pdf")
+ }
+ case _ if p == Programs.MeanQualityByCycle.toString => {
+ outputFiles :+= new File(outputName + ".quality_by_cycle_metrics")
+ outputFiles :+= new File(outputName + ".quality_by_cycle.pdf")
+ }
+ case _ if p == Programs.CollectBaseDistributionByCycle.toString => {
+ outputFiles :+= new File(outputName + ".base_distribution_by_cycle_metrics")
+ outputFiles :+= new File(outputName + ".base_distribution_by_cycle.pdf")
+ }
+ case _ => BiopetQScript.addError("Program '" + p + "' does not exist for 'CollectMultipleMetrics'")
+ })
+ }
+
+ override def commandLine = super.commandLine +
+ required("INPUT=", input, spaceSeparated = false) +
+ required("OUTPUT=", outputName, spaceSeparated = false) +
+ required("ASSUME_SORTED=", assumeSorted, spaceSeparated = false) +
+ required("STOP_AFTER=", stopAfter, spaceSeparated = false) +
+ repeat("PROGRAM=", program, spaceSeparated = false)
+}
+
+object CollectMultipleMetrics {
+ object Programs extends Enumeration {
+ val CollectAlignmentSummaryMetrics, CollectInsertSizeMetrics, QualityScoreDistribution, MeanQualityByCycle, CollectBaseDistributionByCycle = Value
+ }
+}
\ No newline at end of file
--
GitLab