Merge branch 'feature-fastq_filter' into 'develop'

Fix for BIOPET-402



See merge request !467

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

@@ -31,6 +31,7 @@ object BiopetToolsExecutable extends BiopetExecutable {
     nl.lumc.sasc.biopet.tools.ExtractAlignedFastq,
     nl.lumc.sasc.biopet.tools.FastqSplitter,
     nl.lumc.sasc.biopet.tools.FastqSync,
+    nl.lumc.sasc.biopet.tools.FastqFilter,
     nl.lumc.sasc.biopet.tools.FindRepeatsPacBio,
     nl.lumc.sasc.biopet.tools.FindOverlapMatch,
     nl.lumc.sasc.biopet.tools.GvcfToBed,

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

+package nl.lumc.sasc.biopet.tools
+
+import java.io.File
+
+import htsjdk.samtools.fastq.{ AsyncFastqWriter, BasicFastqWriter, FastqReader }
+import nl.lumc.sasc.biopet.utils.ToolCommand
+
+import scala.util.matching.Regex
+import scala.collection.JavaConversions._
+
+/**
+ * Created by pjvan_thof on 28-10-16.
+ */
+object FastqFilter extends ToolCommand {
+  /**
+   * Arg for commandline program
+   * @param inputFile input fastq file
+   * @param outputFile output fastq files
+   */
+  case class Args(inputFile: File = null,
+                  outputFile: File = null,
+                  idRegex: Option[Regex] = None) extends AbstractArgs
+
+  class OptParser extends AbstractOptParser {
+    opt[File]('I', "inputFile") required () valueName "<file>" action { (x, c) =>
+      c.copy(inputFile = x)
+    } text "Path to input file"
+    opt[File]('o', "output") required () unbounded () valueName "<file>" action { (x, c) =>
+      c.copy(outputFile = x)
+    } text "Path to output file"
+    opt[String]("idRegex") unbounded () valueName "<file>" action { (x, c) =>
+      c.copy(idRegex = Some(x.r))
+    } text "Regex to match ID"
+  }
+
+  def main(args: Array[String]): Unit = {
+    val argsParser = new OptParser
+    val cmdArgs: Args = argsParser.parse(args, Args()) getOrElse (throw new IllegalArgumentException)
+
+    logger.info("Start")
+
+    val reader = new FastqReader(cmdArgs.inputFile)
+    val writer = new AsyncFastqWriter(new BasicFastqWriter(cmdArgs.outputFile), 10000)
+
+    var total = 0
+    var kept = 0
+    for (record <- reader.iterator()) {
+      if (cmdArgs.idRegex.map(_.findFirstIn(record.getReadHeader.takeWhile(_ != ' ')).isDefined).getOrElse(true)) {
+        writer.write(record)
+        kept += 1
+      }
+      total += 1
+      if (total % 100000 == 0) logger.info(s"Total reads: $total,  reads left: $kept")
+    }
+    logger.info(s"Total reads: $total,  reads left: $kept")
+
+    writer.close()
+    reader.close()
+
+    logger.info("Done")
+  }
+}

biopet-tools/src/test/resources/paired01_post_filter.fq

+@r01_filter hello
+A
++
+H
+@r03_filter
+G
++
+H

biopet-tools/src/test/resources/paired01_pre_filter.fq

+@r01_filter hello
+A
++
+H
+@r02
+T
++
+I
+@r03_filter
+G
++
+H
+@r04
+C
++
+I
+@r05
+A
++
+H

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

+/**
+ * Biopet is built on top of GATK Queue for building bioinformatic
+ * pipelines. It is mainly intended to support LUMC SHARK cluster which is running
+ * SGE. But other types of HPC that are supported by GATK Queue (such as PBS)
+ * should also be able to execute Biopet tools and pipelines.
+ *
+ * Copyright 2014 Sequencing Analysis Support Core - Leiden University Medical Center
+ *
+ * Contact us at: sasc@lumc.nl
+ *
+ * A dual licensing mode is applied. The source code within this project is freely available for non-commercial use under an AGPL
+ * license; For commercial users or users who do not want to follow the AGPL
+ * license, please contact us to obtain a separate license.
+ */
+package nl.lumc.sasc.biopet.tools
+
+import java.io.File
+import java.nio.file.Paths
+
+import org.scalatest.Matchers
+import org.scalatest.mock.MockitoSugar
+import org.scalatest.testng.TestNGSuite
+import org.testng.annotations.Test
+
+import scala.io.Source
+
+/**
+ * Created by ahbbollen on 27-8-15.
+ */
+class FastqFilterTest extends TestNGSuite with MockitoSugar with Matchers {
+
+  import FastqFilter._
+  private def resourcePath(p: String): String = {
+    Paths.get(getClass.getResource(p).toURI).toString
+  }
+
+  val preFilterFastq = resourcePath("/paired01_pre_filter.fq")
+  val postFilterFastq = resourcePath("/paired01_post_filter.fq")
+
+  @Test
+  def testMain() = {
+    val temp = File.createTempFile("out", ".fastq")
+    temp.deleteOnExit()
+    val args = Array("-I", preFilterFastq, "-o", temp.getAbsolutePath, "--idRegex", "_filter$")
+    main(args)
+
+    Source.fromFile(temp).getLines().toList shouldBe Source.fromFile(postFilterFastq).getLines().toList
+  }
+
+}