From 798fb6a7a23cd110d3a82fe0d227fbde074c16fe Mon Sep 17 00:00:00 2001
From: hanna <hanna@348d0f76-0448-11de-a6fe-93d51630548a>
Date: Mon, 4 Apr 2011 15:35:14 +0000
Subject: [PATCH] First draft of a script to measure performance of read
 walkers when merging dynamically.

git-svn-id: file:///humgen/gsa-scr1/gsa-engineering/svn_contents/trunk@5570 348d0f76-0448-11de-a6fe-93d51630548a
---
 .../hanna/PrintReadsAcrossManySamples.q       | 81 +++++++++++++++++++
 scala/qscript/oneoffs/hanna/runprintreads.sh  |  6 ++
 2 files changed, 87 insertions(+)
 create mode 100644 scala/qscript/oneoffs/hanna/PrintReadsAcrossManySamples.q
 create mode 100644 scala/qscript/oneoffs/hanna/runprintreads.sh

diff --git a/scala/qscript/oneoffs/hanna/PrintReadsAcrossManySamples.q b/scala/qscript/oneoffs/hanna/PrintReadsAcrossManySamples.q
new file mode 100644
index 000000000..98769e5df
--- /dev/null
+++ b/scala/qscript/oneoffs/hanna/PrintReadsAcrossManySamples.q
@@ -0,0 +1,81 @@
+import java.io.PrintWriter
+import org.broadinstitute.sting.queue.QScript
+import org.broadinstitute.sting.queue.extensions.gatk._
+import org.broadinstitute.sting.utils.exceptions.ReviewedStingException
+
+/**
+ * A pipeline for Queue that runs a custom walker outside of the GATK jar.
+ * NOTE: This code is an unsupported example for soliciting feedback on how to improve Queue.
+ * Future syntax will simplify running the GATK so please expect the syntax below to change significantly.
+ */
+class PrintReadsAcrossManySamples extends QScript {
+  // The full packaged jar should be used.
+  // You can build this jar via 'ant package' and then find it under
+  // 'Sting/dist/packages/GenomeAnalysisTK-*/GenomeAnalysisTK.jar'
+  @Input(doc="The path to the packaged GenomeAnalysisTK.jar file.", shortName="gatk")
+  var gatkJar: File = null
+
+  @Input(doc="The reference file for the bam files.", shortName="R")
+  var referenceFile: File = null
+
+  // NOTE: Do not initialize List, Set, or Option to null
+  // as you won't be able to update the collection.
+  // By default set:
+  //   List[T] = Nil
+  //   Set[T] = Set.empty[T]
+  //   Option[T] = None
+  @Input(doc="One or more bam files.", shortName="I")
+  var bamFiles: List[File] = Nil
+
+  @Input(doc="Name of the test case", fullName="test_case",required=false)
+  var testCaseName: String = "."
+
+  @Input(doc="Max number of bam files to process", fullName="max_bams",required=false)
+  var maxBams = 1
+
+  @Input(doc="Step size",fullName="step_size",required=false)
+  var stepSize = 1
+
+  // This trait allows us set the variables below in one place,
+  // and then reuse this trait on each CommandLineGATK function below.
+  trait PrintReadsAcrossManySamplesArguments extends CommandLineGATK {
+    this.jarFile = PrintReadsAcrossManySamples.this.gatkJar
+    this.reference_sequence = PrintReadsAcrossManySamples.this.referenceFile
+    this.memoryLimit = 8
+  }
+
+
+  def script = {
+    if(bamFiles.size != 1)
+      throw new ReviewedStingException("-I argument must consist of exactly one file containing a list of BAM files.");
+
+    var lines: List[String] = List[String]()
+    for(line <- scala.io.Source.fromFile(bamFiles(0)).getLines) {
+      lines = lines ::: List(line)
+    }
+
+    for(numBams <- 1 to math.min(maxBams,lines.size) by stepSize) {
+      val dir = new File(testCaseName + "/%03d_bams".format(numBams))
+      dir.mkdir()
+
+      val file = new File(dir,"bams.list")
+      val writer = new PrintWriter(file)
+      
+      for(bamIndex <- 0 to numBams-1)
+        writer.println(lines(bamIndex))
+
+      writer.close()
+
+      // Create the function that we can run.
+      val printreads = new PrintReads with PrintReadsAcrossManySamplesArguments
+
+      printreads.jobOutputFile = new File(dir, "PrintReads.out")      
+      printreads.input_file = List(file)
+      printreads.reference_sequence = referenceFile
+      printreads.out = new File("/dev/null")
+
+      add(printreads)
+    }
+
+  }
+}
diff --git a/scala/qscript/oneoffs/hanna/runprintreads.sh b/scala/qscript/oneoffs/hanna/runprintreads.sh
new file mode 100644
index 000000000..daf395ed6
--- /dev/null
+++ b/scala/qscript/oneoffs/hanna/runprintreads.sh
@@ -0,0 +1,6 @@
+#!/bin/sh
+java -Djava.io.tmpdir=/broad/shptmp/hanna -jar ~/src/Sting/dist/Queue.jar \
+    --script PrintReadsAcrossManySamples.q \
+    -gatk ~/src/Sting/dist/GenomeAnalysisTK.jar \
+    -R /humgen/1kg/reference/human_g1k_v37.fasta \
+    -I ~/tests/1600samples/1kg_t2d.list --max_bams 2000 --step_size 10 -bsub -jobQueue week $1