Added insertsize plot

public/bammetrics/src/main/scala/nl/lumc/sasc/biopet/pipelines/bammetrics/BammetricsReport.scala

@@ -4,7 +4,7 @@ import java.io.{ PrintWriter, File }
 
 import nl.lumc.sasc.biopet.core.report.{ ReportBuilder, ReportPage, ReportSection }
 import nl.lumc.sasc.biopet.core.summary.{ SummaryValue, Summary }
-import nl.lumc.sasc.biopet.extensions.rscript.StackedBarPlot
+import nl.lumc.sasc.biopet.extensions.rscript.{XYPlot, StackedBarPlot}
 
 /**
  * Created by pjvan_thof on 3/30/15.
@@ -83,9 +83,68 @@ object BammetricsReport extends ReportBuilder {
     plot.input = tsvFile
     plot.output = pngFile
     plot.ylabel = Some("Reads")
-    plot.width = Some(750)
+    plot.width = Some(1200)
     plot.title = Some("Aligned reads")
     plot.runLocal()
   }
 
+  def insertSizePlot(outputDir: File,
+                           prefix: String,
+                           summary: Summary,
+                           libraryLevel: Boolean = false,
+                           sampleId: Option[String] = None): Unit = {
+    val tsvFile = new File(outputDir, prefix + ".tsv")
+    val pngFile = new File(outputDir, prefix + ".png")
+    val tsvWriter = new PrintWriter(tsvFile)
+    if (libraryLevel) {
+      tsvWriter.print("Library")
+    } else {
+      sampleId match {
+        case Some(sample) => tsvWriter.println("\t" + sample)
+        case _ => tsvWriter.println(summary.samples.mkString("Sample\t","\t", ""))
+      }
+    }
+
+    var map: Map[Int, Map[String, Int]] = Map()
+
+    def fill(sample: String, lib: Option[String]): Unit = {
+      new SummaryValue(List("bammetrics", "stats", "CollectInsertSizeMetrics", "histogram", "content"),
+        summary, Some(sample), lib).value.getOrElse(List(List("insert_size","All_Reads.fr_count"))) match {
+        case l:List[_] => {
+          l.tail.foreach(_ match {
+            case l:List[_] => {
+              val insertSize = l.head.toString.toInt
+              val count = l.tail.head.toString.toInt
+              val old = map.getOrElse(insertSize, Map())
+              map += insertSize -> (old + ((sample + lib.getOrElse("")) -> count))
+            }
+            case _ => throw new IllegalStateException("Must be a list")
+          })
+        }
+        case _ => throw new IllegalStateException("Must be a list")
+      }
+    }
+
+    summary.samples.foreach(fill(_, None))
+
+    for ((insertSize, counts) <- map) {
+      tsvWriter.print(insertSize)
+      for (sample <- summary.samples) {
+        tsvWriter.print("\t" + counts.getOrElse(sample, "0"))
+      }
+      tsvWriter.println()
+    }
+
+    tsvWriter.close()
+
+    val plot = new XYPlot(null)
+    plot.input = tsvFile
+    plot.output = pngFile
+    plot.ylabel = Some("Reads")
+    plot.xlabel = Some("Insertsize")
+    plot.width = Some(1200)
+    plot.title = Some("Insertsize")
+    plot.runLocal()
+
+  }
 }

public/biopet-framework/src/main/resources/nl/lumc/sasc/biopet/extensions/rscript/plotXY.R

+library(reshape2)
+library(ggplot2)
+library(argparse)
+
+parser <- ArgumentParser(description='Process some integers')
+parser$add_argument('--input', dest='input', type='character', help='Input tsv file', required=TRUE)
+parser$add_argument('--output', dest='output', type='character', help='Output png file', required=TRUE)
+parser$add_argument('--width', dest='width', type='integer', default = 500)
+parser$add_argument('--height', dest='height', type='integer', default = 500)
+parser$add_argument('--xlabel', dest='xlabel', type='character')
+parser$add_argument('--ylabel', dest='ylabel', type='character', required=TRUE)
+parser$add_argument('--llabel', dest='llabel', type='character')
+parser$add_argument('--title', dest='title', type='character')
+
+arguments <- parser$parse_args()
+
+png(filename = arguments$output, width = arguments$width, height = arguments$height)
+
+DF <- read.table(arguments$input, header=TRUE)
+
+if (is.null(arguments$xlabel)) xlab <- colnames(DF)[1] else xlab <- arguments$xlabel
+
+colnames(DF)[1] <- "Rank"
+
+DF1 <- melt(DF, id.var="Rank")
+
+DF1 <- DF1[DF1$value > 0, ]
+
+ggplot(DF1, aes(x = Rank, y = value, group = variable, color = variable)) +
+  xlab(xlab) +
+  ylab(arguments$ylabel) +
+  guides(fill=guide_legend(title=arguments$llabel)) +
+  theme(axis.text.x = element_text(angle = 90, hjust = 1, size = 8)) +
+  ggtitle(arguments$title) +
+  geom_line()
+
+dev.off()

public/biopet-framework/src/main/resources/nl/lumc/sasc/biopet/tools/plotXY.R

-library('ggplot2')
-library('reshape2')
-
-args <- commandArgs(TRUE)
-inputArg <- args[1]
-outputArg <- args[2]
-
-tsv<-read.table(inputArg, header = 1, sep= '\t', stringsAsFactors = F)
-
-data <- melt(tsv)
-
-data$X <- as.numeric(data$X)
-data <- na.omit(data)
-data <- data[data$value > 0,]
-
-print("Starting to plot")
-png(file = outputArg, width = 1500, height = 1500)
-ggplot(data, aes(x=X, y=value, color=variable, group=variable)) + geom_line()
-dev.off()
-print("plot done")

public/biopet-framework/src/main/scala/nl/lumc/sasc/biopet/extensions/rscript/XYPlot.scala

+package nl.lumc.sasc.biopet.extensions.rscript
+
+import java.io.File
+
+import nl.lumc.sasc.biopet.core.config.Configurable
+import nl.lumc.sasc.biopet.extensions.RscriptCommandLineFunction
+import org.broadinstitute.gatk.utils.commandline.{Input, Output}
+
+/**
+ * Created by pjvan_thof on 4/29/15.
+ */
+class XYPlot(val root: Configurable) extends RscriptCommandLineFunction {
+  protected var script: File = config("script", default = "plotXY.R")
+
+  @Input
+  var input: File = _
+
+  @Output
+  var output: File = _
+
+  var width: Option[Int] = config("width")
+  var height: Option[Int] = config("height")
+  var xlabel: Option[String] = config("xlabel")
+  var ylabel: Option[String] = config("ylabel")
+  var llabel: Option[String] = config("llabel")
+  var title: Option[String] = config("title")
+
+  override def cmdLine: String = super.cmdLine +
+    required("--input", input) +
+    required("--output", output) +
+    optional("--width", width) +
+    optional("--height", height) +
+    optional("--xlabel", xlabel) +
+    required("--ylabel", ylabel) +
+    optional("--llabel", llabel) +
+    optional("--title", title)
+}