Merge branch 'fix-278' into 'develop'

Added fix mpileup for bcftools

fixes #278 

See merge request !326

public/biopet-core/src/main/scala/nl/lumc/sasc/biopet/core/extensions/PythonCommandLineFunction.scala

@@ -24,7 +24,7 @@ trait PythonCommandLineFunction extends BiopetCommandLineFunction {
   @Input(doc = "Python script", required = false)
   var python_script: File = _
 
-  executable = config("exe", default = "python", submodule = "python")
+  executable = config("exe", default = "python", submodule = "python", freeVar = false)
 
   protected var python_script_name: String = _
 

public/biopet-extensions/src/main/resources/nl/lumc/sasc/biopet/extensions/samtools/fix_iupac_mpileup.py

+#!/usr/bin/env python
+#
+# 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 that are
+# not part of GATK Queue 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.
+#
+
+
+from __future__ import print_function
+
+__author__="Peter van 't Hof"
+
+import sys
+import re
+
+upacPatern = re.compile(r'[RYKMSWBDHV]')
+
+if __name__ == "__main__":
+    for line in sys.stdin:
+        l = line.strip().split("\t")
+        if len(l) >= 3:
+            l[3] = upacPatern.sub("N", l[3])
+
+        print("\t".join(map(str, l)))

public/biopet-extensions/src/main/resources/nl/lumc/sasc/biopet/extensions/varscan/fix_mpileup.py

@@ -33,7 +33,9 @@ if __name__ == "__main__":
     """
     for line in sys.stdin:
         l = line.strip().split("\t")
-        if l[3] == "0":
+        l[2] = upacPatern.sub("N", l[2])
+
+        if len(l) < 4 or l[3] == "0":
             # no alignment to this position
             print("\t".join(map(str, l)))
             continue
@@ -49,5 +51,4 @@ if __name__ == "__main__":
         if new_size == 0:
             l[5] = ""
 
-        l[2] = upacPatern.sub("N", l[2])
         print("\t".join(map(str, l)))

public/biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/samtools/FixMpileup.scala

+package nl.lumc.sasc.biopet.extensions.samtools
+
+import nl.lumc.sasc.biopet.core.extensions.PythonCommandLineFunction
+import nl.lumc.sasc.biopet.utils.config.Configurable
+
+/**
+ * Created by sajvanderzeeuw on 19-1-16.
+ */
+class FixMpileup(val root: Configurable) extends PythonCommandLineFunction {
+  setPythonScript("fix_iupac_mpileup.py", "/nl/lumc/sasc/biopet/extensions/samtools/")
+  def cmdLine = getPythonCommand
+}

public/biopet-extensions/src/main/scala/nl/lumc/sasc/biopet/extensions/samtools/SamtoolsMpileup.scala

@@ -37,6 +37,7 @@ class SamtoolsMpileup(val root: Configurable) extends Samtools with Reference {
 
   var disableBaq: Boolean = config("disable_baq", default = false)
   var u: Boolean = config("u", default = false)
+  var v: Boolean = config("u", default = false)
   var minMapQuality: Option[Int] = config("min_map_quality")
   var minBaseQuality: Option[Int] = config("min_base_quality")
   var depth: Option[Int] = config("depth")
@@ -57,6 +58,7 @@ class SamtoolsMpileup(val root: Configurable) extends Samtools with Reference {
     conditional(outputMappingQuality, "-s") +
     conditional(disableBaq, "-B") +
     conditional(u, "-u") +
+    conditional(v, "-v") +
     (if (outputAsStsout) "" else required("-o", output)) +
     (if (inputAsStdin) "-" else repeat(input))
 }

public/shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/Bcftools.scala

@@ -2,7 +2,7 @@ package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
 import nl.lumc.sasc.biopet.extensions.Tabix
 import nl.lumc.sasc.biopet.extensions.bcftools.BcftoolsCall
-import nl.lumc.sasc.biopet.extensions.samtools.SamtoolsMpileup
+import nl.lumc.sasc.biopet.extensions.samtools.{ FixMpileup, SamtoolsMpileup }
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** default mode of bcftools */
@@ -14,6 +14,7 @@ class Bcftools(val root: Configurable) extends Variantcaller {
     val mp = new SamtoolsMpileup(this)
     mp.input = inputBams.values.toList
     mp.u = true
+    mp.v = true
     mp.reference = referenceFasta()
 
     val bt = new BcftoolsCall(this)
@@ -21,7 +22,7 @@ class Bcftools(val root: Configurable) extends Variantcaller {
     bt.v = true
     bt.c = true
 
-    add(mp | bt > outputFile)
+    add(mp | new FixMpileup(this) | bt > outputFile)
     add(Tabix(this, outputFile))
   }
 }

public/shiva/src/main/scala/nl/lumc/sasc/biopet/pipelines/shiva/variantcallers/BcftoolsSingleSample.scala

 package nl.lumc.sasc.biopet.pipelines.shiva.variantcallers
 
-import java.io.File
-
 import nl.lumc.sasc.biopet.extensions.{ Ln, Tabix }
 import nl.lumc.sasc.biopet.extensions.bcftools.{ BcftoolsMerge, BcftoolsCall }
-import nl.lumc.sasc.biopet.extensions.samtools.SamtoolsMpileup
+import nl.lumc.sasc.biopet.extensions.samtools.{ FixMpileup, SamtoolsMpileup }
 import nl.lumc.sasc.biopet.utils.config.Configurable
 
 /** default mode of bcftools */
@@ -17,6 +15,7 @@ class BcftoolsSingleSample(val root: Configurable) extends Variantcaller {
       val mp = new SamtoolsMpileup(this)
       mp.input :+= inputBam
       mp.u = true
+      mp.v = true
       mp.reference = referenceFasta()
 
       val bt = new BcftoolsCall(this)
@@ -25,7 +24,7 @@ class BcftoolsSingleSample(val root: Configurable) extends Variantcaller {
       bt.c = true
       bt.output = new File(outputDir, sample + ".vcf.gz")
 
-      add(mp | bt)
+      add(mp | new FixMpileup(this) | bt)
       add(Tabix(this, bt.output))
       bt.output
     }