Merge branch 'feature-find_overlap' into 'develop'

Feature find overlap

This is a sample tool to find overlaping samples based on a matrix produces by vcfstats

See merge request !454

biopet-tools-package/src/main/scala/nl/lumc/sasc/biopet/BiopetToolsExecutable.scala

@@ -32,6 +32,7 @@ object BiopetToolsExecutable extends BiopetExecutable {
     nl.lumc.sasc.biopet.tools.FastqSplitter,
     nl.lumc.sasc.biopet.tools.FastqSync,
     nl.lumc.sasc.biopet.tools.FindRepeatsPacBio,
+    nl.lumc.sasc.biopet.tools.FindOverlapMatch,
     nl.lumc.sasc.biopet.tools.GvcfToBed,
     nl.lumc.sasc.biopet.tools.MergeAlleles,
     nl.lumc.sasc.biopet.tools.MergeTables,

biopet-tools/src/main/scala/nl/lumc/sasc/biopet/tools/FindOverlapMatch.scala

+package nl.lumc.sasc.biopet.tools
+
+import java.io.{ File, PrintStream }
+
+import nl.lumc.sasc.biopet.utils.ToolCommand
+
+import scala.collection.mutable.ListBuffer
+import scala.io.Source
+
+/**
+ * This tool will find all pairs above a cutoff in a data table
+ *
+ * Created by pjvan_thof on 21-9-16.
+ */
+object FindOverlapMatch extends ToolCommand {
+
+  case class Args(inputMetrics: File = null,
+                  outputFile: Option[File] = None,
+                  cutoff: Double = 0.0,
+                  filterSameNames: Boolean = true) extends AbstractArgs
+
+  class OptParser extends AbstractOptParser {
+    opt[File]('i', "input") required () unbounded () valueName "<file>" action { (x, c) =>
+      c.copy(inputMetrics = x)
+    } text "Input should be a table where the first row and column have the ID's, those can be different"
+    opt[File]('o', "output") unbounded () valueName "<file>" action { (x, c) =>
+      c.copy(outputFile = Some(x))
+    } text "default to stdout"
+    opt[Double]('c', "cutoff") required () unbounded () valueName "<value>" action { (x, c) =>
+      c.copy(cutoff = x)
+    } text "minimum value to report it as pair"
+    opt[Unit]("use_same_names") unbounded () valueName "<value>" action { (x, c) =>
+      c.copy(filterSameNames = false)
+    } text "Do not compare samples with the same name"
+
+  }
+
+  /**
+   * @param args the command line arguments
+   */
+  def main(args: Array[String]): Unit = {
+    val argsParser = new OptParser
+    val cmdArgs: Args = argsParser.parse(args, Args()) getOrElse (throw new IllegalArgumentException)
+
+    val reader = Source.fromFile(cmdArgs.inputMetrics)
+
+    val data = reader.getLines().map(_.split("\t")).toArray
+
+    val samplesColumnHeader = data.head.zipWithIndex.tail
+    val samplesRowHeader = data.map(_.head).zipWithIndex.tail
+
+    var overlap = 0
+    var multiOverlap = 0
+    var noOverlap = 0
+
+    val writer = cmdArgs.outputFile match {
+      case Some(file) => new PrintStream(file)
+      case _          => sys.process.stdout
+    }
+
+    for (i1 <- samplesColumnHeader) {
+      val buffer = ListBuffer[(String, Double)]()
+      for (i2 <- samplesRowHeader) {
+        val value = data(i1._2)(i2._2).toDouble
+        if (value >= cmdArgs.cutoff && (!cmdArgs.filterSameNames || i1._2 != i2._2)) {
+          buffer.+=((i2._1, value))
+        }
+      }
+      if (buffer.nonEmpty) overlap += 1
+      else noOverlap += 1
+      if (buffer.size > 1) multiOverlap += 1
+
+      writer.println(s"${i1._1}\t${buffer.mkString("\t")}")
+    }
+    logger.info(s"$overlap found")
+    logger.info(s"no $noOverlap found")
+    logger.info(s"multi $multiOverlap found")
+    writer.close()
+  }
+}

biopet-tools/src/test/resources/overlapmetrics.default.output

+sample1	(sample3,0.9)
+sample2	
+sample3	(sample1,0.9)

biopet-tools/src/test/resources/overlapmetrics.same_names.output

+sample1	(sample1,1.0)	(sample3,0.9)
+sample2	(sample2,1.0)
+sample3	(sample1,0.9)	(sample3,1.0)

biopet-tools/src/test/resources/overlapmetrics.txt

+	sample1	sample2	sample3
+sample1	1.0	0.5	0.9
+sample2	0.5	1.0	0.5
+sample3	0.9	0.5	1.0

biopet-tools/src/test/scala/nl/lumc/sasc/biopet/tools/FindOverlapMatchTest.scala

+package nl.lumc.sasc.biopet.tools
+
+import java.io.File
+import java.nio.file.Paths
+
+import org.scalatest.Matchers
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.io.Source
+
+/**
+ * Created by pjvan_thof on 27-9-16.
+ */
+class FindOverlapMatchTest extends TestNGSuite with Matchers {
+
+  private def resourcePath(p: String): String = {
+    Paths.get(getClass.getResource(p).toURI).toString
+  }
+
+  @Test
+  def testOverlap: Unit = {
+    val input = new File(resourcePath("/overlapmetrics.txt"))
+    val output = File.createTempFile("overlap.", ".txt")
+    val shouldBeOutput = new File(resourcePath("/overlapmetrics.default.output"))
+    output.deleteOnExit()
+    FindOverlapMatch.main(Array("-i", input.getAbsolutePath, "-c", "0.9", "-o", output.getAbsolutePath))
+    Source.fromFile(output).getLines().toList shouldBe Source.fromFile(shouldBeOutput).getLines().toList
+  }
+
+  @Test
+  def testOverlapSameName: Unit = {
+    val input = new File(resourcePath("/overlapmetrics.txt"))
+    val output = File.createTempFile("overlap.", ".txt")
+    val shouldBeOutput = new File(resourcePath("/overlapmetrics.same_names.output"))
+    output.deleteOnExit()
+    FindOverlapMatch.main(Array("-i", input.getAbsolutePath, "-c", "0.9", "-o", output.getAbsolutePath, "--use_same_names"))
+    Source.fromFile(output).getLines().toList shouldBe Source.fromFile(shouldBeOutput).getLines().toList
+  }
+
+}