Code polishing

annotate.nf

@@ -351,6 +351,7 @@ def sarekMessage() {
 
 def startMessage() {
   // Display start message
+  SarekUtils.sarek_ascii()
   this.sarekMessage()
   this.minimalInformationMessage()
 }

buildContainers.nf

@@ -248,6 +248,7 @@ def sarekMessage() {
 
 def startMessage() {
   // Display start message
+  SarekUtils.sarek_ascii()
   this.sarekMessage()
   this.minimalInformationMessage()
 }

buildReferences.nf

@@ -296,6 +296,7 @@ def sarekMessage() {
 
 def startMessage() {
   // Display start message
+  SarekUtils.sarek_ascii()
   this.sarekMessage()
   this.minimalInformationMessage()
 }

germlineVC.nf

@@ -65,8 +65,8 @@ directoryMap = SarekUtils.defineDirectoryMap(params.outDir)
 referenceMap = defineReferenceMap()
 toolList = defineToolList()
 
-if (!checkReferenceMap(referenceMap)) exit 1, 'Missing Reference file(s), see --help for more information'
-if (!checkParameterList(tools,toolList)) exit 1, 'Unknown tool(s), see --help for more information'
+if (!SarekUtils.checkReferenceMap(referenceMap)) exit 1, 'Missing Reference file(s), see --help for more information'
+if (!SarekUtils.checkParameterList(tools,toolList)) exit 1, 'Unknown tool(s), see --help for more information'
 
 if (params.test && params.genome in ['GRCh37', 'GRCh38']) {
   referenceMap.intervals = file("$workflow.projectDir/repeats/tiny_${params.genome}.list")
@@ -88,11 +88,9 @@ else tsvPath = "${directoryMap.recalibrated}/recalibrated.tsv"
 bamFiles = Channel.empty()
 if (tsvPath) {
   tsvFile = file(tsvPath)
-  bamFiles = extractBams(tsvFile)
+  bamFiles = SarekUtils.extractBams(tsvFile, "germline")
 } else exit 1, 'No sample were defined, see --help'
 
-(patientGenders, bamFiles) = extractGenders(bamFiles)
-
 /*
 ================================================================================
 =                               P R O C E S S E S                              =
@@ -273,9 +271,7 @@ recalTables = recalTables
     [patient, sample, bam, bai, intervalBed, recalTable] }
 
 // re-associate the BAMs and samples with the recalibration table
-bamsForHC = bamsForHC
-  .phase(recalTables) { it[0..4] }
-  .map { it1, it2 -> it1 + [it2[6]] }
+bamsForHC = bamsForHC.join(recalTables, by:[0,1,2,3,4])
 
 bamsAll = bamsNormal.combine(bamsTumor)
 
@@ -643,50 +639,11 @@ process GetVersionVCFtools {
 ================================================================================
 */
 
-def checkFileExtension(it, extension) {
-  // Check file extension
-  if (!it.toString().toLowerCase().endsWith(extension.toLowerCase())) exit 1, "File: ${it} has the wrong extension: ${extension} see --help for more information"
-}
-
-def checkParameterExistence(it, list) {
-  // Check parameter existence
-  if (!list.contains(it)) {
-    println("Unknown parameter: ${it}")
-    return false
-  }
-  return true
-}
-
-def checkParameterList(list, realList) {
-  // Loop through all parameters to check their existence and spelling
-  return list.every{ checkParameterExistence(it, realList) }
-}
-
 def checkParamReturnFile(item) {
   params."${item}" = params.genomes[params.genome]."${item}"
   return file(params."${item}")
 }
 
-def checkReferenceMap(referenceMap) {
-  // Loop through all the references files to check their existence
-  referenceMap.every {
-    referenceFile, fileToCheck ->
-    checkRefExistence(referenceFile, fileToCheck)
-  }
-}
-
-def checkRefExistence(referenceFile, fileToCheck) {
-  if (fileToCheck instanceof List) return fileToCheck.every{ checkRefExistence(referenceFile, it) }
-  def f = file(fileToCheck)
-  // this is an expanded wildcard: we can assume all files exist
-  if (f instanceof List && f.size() > 0) return true
-  else if (!f.exists()) {
-    log.info  "Missing references: ${referenceFile} ${fileToCheck}"
-    return false
-  }
-  return true
-}
-
 def checkUppmaxProject() {
   // check if UPPMAX project number is specified
   return !(workflow.profile == 'slurm' && !params.project)
@@ -720,39 +677,6 @@ def defineToolList() {
   ]
 }
 
-def extractBams(tsvFile) {
-  // Channeling the TSV file containing BAM.
-  // Format is: "subject gender status sample bam bai"
-  Channel
-    .from(tsvFile.readLines())
-    .map{line ->
-      def list      = returnTSV(line.split(),6)
-      def idPatient = list[0]
-      def gender    = list[1]
-      def status    = returnStatus(list[2].toInteger())
-      def idSample  = list[3]
-      def bamFile   = returnFile(list[4])
-      def baiFile   = returnFile(list[5])
-
-      checkFileExtension(bamFile,".bam")
-      checkFileExtension(baiFile,".bai")
-
-      [ idPatient, gender, status, idSample, bamFile, baiFile ]
-    }
-}
-
-def extractGenders(channel) {
-  def genders = [:]  // an empty map
-  channel = channel.map{ it ->
-    def idPatient = it[0]
-    def gender = it[1]
-    genders[idPatient] = gender
-
-    [idPatient] + it[2..-1]
-  }
-  [genders, channel]
-}
-
 def generateIntervalsForVC(bams, intervals) {
   def (bamsNew, bamsForVC) = bams.into(2)
   def (intervalsNew, vcIntervals) = intervals.into(2)
@@ -826,34 +750,14 @@ def nextflowMessage() {
   log.info "N E X T F L O W  ~  version ${workflow.nextflow.version} ${workflow.nextflow.build}"
 }
 
-def returnFile(it) {
-  // return file if it exists
-  if (!file(it).exists()) exit 1, "Missing file in TSV file: ${it}, see --help for more information"
-  return file(it)
-}
-
-def returnStatus(it) {
-  // Return status if it's correct
-  // Status should be only 0 or 1
-  // 0 being normal
-  // 1 being tumor (or relapse or anything that is not normal...)
-  if (!(it in [0, 1])) exit 1, "Status is not recognized in TSV file: ${it}, see --help for more information"
-  return it
-}
-
-def returnTSV(it, number) {
-  // return TSV if it has the correct number of items in row
-  if (it.size() != number) exit 1, "Malformed row in TSV file: ${it}, see --help for more information"
-  return it
-}
-
 def sarekMessage() {
   // Display Sarek message
   log.info "Sarek - Workflow For Somatic And Germline Variations ~ ${params.version} - " + this.grabRevision() + (workflow.commitId ? " [${workflow.commitId}]" : "")
 }
 
 def startMessage() {
   // Display start message
+  SarekUtils.sarek_ascii()
   this.sarekMessage()
   this.minimalInformationMessage()
 }

lib/SarekUtils.groovy

@@ -1,5 +1,28 @@
-class MyUtils {
-  // Check if params is in this given list
+import static nextflow.Nextflow.file
+import nextflow.Channel
+
+class SarekUtils {
+
+  // Check file extension
+  static def checkFileExtension(it, extension) {
+    if (!it.toString().toLowerCase().endsWith(extension.toLowerCase())) exit 1, "File: ${it} has the wrong extension: ${extension} see --help for more information"
+  }
+
+  // Check parameter existence
+  static def checkParameterExistence(it, list) {
+    if (!list.contains(it)) {
+      println("Unknown parameter: ${it}")
+      return false
+    }
+    return true
+  }
+
+  // Compare each parameter with a list of parameters
+  static def checkParameterList(list, realList) {
+    return list.every{ checkParameterExistence(it, realList) }
+  }
+
+  // Return element in list of allowed params
   static def checkParams(it) {
     return it in [
       'annotate-tools',
@@ -68,15 +91,22 @@ class MyUtils {
       'version']
   }
 
-  // Loop through all parameters to check their existence and spelling
-  static def checkParameterList(list, realList) {
-    return list.every{ checkParameterExistence(it, realList) }
+  // Loop through all the references files to check their existence
+  static def checkReferenceMap(referenceMap) {
+    referenceMap.every {
+      referenceFile, fileToCheck ->
+      SarekUtils.checkRefExistence(referenceFile, fileToCheck)
+    }
   }
 
-  // Check parameter existence
-  static def checkParameterExistence(it, list) {
-    if (!list.contains(it)) {
-      println("Unknown parameter: ${it}")
+  // Loop through all the references files to check their existence
+  static def checkRefExistence(referenceFile, fileToCheck) {
+    if (fileToCheck instanceof List) return fileToCheck.every{ SarekUtils.checkRefExistence(referenceFile, it) }
+    def f = file(fileToCheck)
+    // this is an expanded wildcard: we can assume all files exist
+    if (f instanceof List && f.size() > 0) return true
+    else if (!f.exists()) {
+      this.log.info  "Missing references: ${referenceFile} ${fileToCheck}"
       return false
     }
     return true
@@ -85,32 +115,66 @@ class MyUtils {
   // Define map of directories
   static def defineDirectoryMap(outDir) {
     return [
-      'nonRealigned'     : "${outDir}/Preprocessing/NonRealigned",
-      'nonRecalibrated'  : "${outDir}/Preprocessing/NonRecalibrated",
-      'recalibrated'     : "${outDir}/Preprocessing/Recalibrated",
-      'ascat'            : "${outDir}/VariantCalling/Ascat",
-      'freebayes'        : "${outDir}/VariantCalling/FreeBayes",
-      'gvcf-hc'          : "${outDir}/VariantCalling/HaplotypeCallerGVCF",
-      'haplotypecaller'  : "${outDir}/VariantCalling/HaplotypeCaller",
-      'manta'            : "${outDir}/VariantCalling/Manta",
-      'mutect1'          : "${outDir}/VariantCalling/MuTect1",
-      'mutect2'          : "${outDir}/VariantCalling/MuTect2",
-      'strelka'          : "${outDir}/VariantCalling/Strelka",
-      'strelkabp'        : "${outDir}/VariantCalling/StrelkaBP",
-      'snpeff'           : "${outDir}/Annotation/SnpEff",
-      'vep'              : "${outDir}/Annotation/VEP",
-      'bamQC'            : "${outDir}/Reports/bamQC",
-      'bcftoolsStats'    : "${outDir}/Reports/BCFToolsStats",
-      'fastQC'           : "${outDir}/Reports/FastQC",
-      'markDuplicatesQC' : "${outDir}/Reports/MarkDuplicates",
-      'multiQC'          : "${outDir}/Reports/MultiQC",
-      'samtoolsStats'    : "${outDir}/Reports/SamToolsStats",
-      'snpeffReports'    : "${outDir}/Reports/SnpEff",
-      'vcftools'         : "${outDir}/Reports/VCFTools",
-      'version'          : "${outDir}/Reports/ToolsVersion"
+    'nonRealigned'     : "${outDir}/Preprocessing/NonRealigned",
+    'nonRecalibrated'  : "${outDir}/Preprocessing/NonRecalibrated",
+    'recalibrated'     : "${outDir}/Preprocessing/Recalibrated",
+    'ascat'            : "${outDir}/VariantCalling/Ascat",
+    'freebayes'        : "${outDir}/VariantCalling/FreeBayes",
+    'gvcf-hc'          : "${outDir}/VariantCalling/HaplotypeCallerGVCF",
+    'haplotypecaller'  : "${outDir}/VariantCalling/HaplotypeCaller",
+    'manta'            : "${outDir}/VariantCalling/Manta",
+    'mutect1'          : "${outDir}/VariantCalling/MuTect1",
+    'mutect2'          : "${outDir}/VariantCalling/MuTect2",
+    'strelka'          : "${outDir}/VariantCalling/Strelka",
+    'strelkabp'        : "${outDir}/VariantCalling/StrelkaBP",
+    'snpeff'           : "${outDir}/Annotation/SnpEff",
+    'vep'              : "${outDir}/Annotation/VEP",
+    'bamQC'            : "${outDir}/Reports/bamQC",
+    'bcftoolsStats'    : "${outDir}/Reports/BCFToolsStats",
+    'fastQC'           : "${outDir}/Reports/FastQC",
+    'markDuplicatesQC' : "${outDir}/Reports/MarkDuplicates",
+    'multiQC'          : "${outDir}/Reports/MultiQC",
+    'samtoolsStats'    : "${outDir}/Reports/SamToolsStats",
+    'snpeffReports'    : "${outDir}/Reports/SnpEff",
+    'vcftools'         : "${outDir}/Reports/VCFTools",
+    'version'          : "${outDir}/Reports/ToolsVersion"
     ]
   }
 
+  // Channeling the TSV file containing BAM.
+  // Format is: "subject gender status sample bam bai"
+  static def extractBams(tsvFile, mode) {
+    Channel
+      .from(tsvFile.readLines())
+      .map{line ->
+        def list      = SarekUtils.returnTSV(line.split(),6)
+        def idPatient = list[0]
+        def gender    = list[1]
+        def status    = SarekUtils.returnStatus(list[2].toInteger())
+        def idSample  = list[3]
+        def bamFile   = SarekUtils.returnFile(list[4])
+        def baiFile   = SarekUtils.returnFile(list[5])
+
+        SarekUtils.checkFileExtension(bamFile,".bam")
+        SarekUtils.checkFileExtension(baiFile,".bai")
+
+        if (mode == "germline") return [ idPatient, status, idSample, bamFile, baiFile ]
+        else return [ idPatient, gender, status, idSample, bamFile, baiFile ]
+      }
+  }
+
+  // Extract gender from Channel as it's only used for CNVs
+  static def extractGenders(channel) {
+    def genders = [:]
+    channel = channel.map{ it ->
+      def idPatient = it[0]
+      def gender = it[1]
+      genders[idPatient] = gender
+      [idPatient] + it[2..-1]
+    }
+    [genders, channel]
+  }
+
   // Compare params to list of verified params
   static def isAllowedParams(params) {
     final test = true
@@ -122,4 +186,34 @@ class MyUtils {
     }
     return test
   }
+
+  // Return file if it exists
+  static def returnFile(it) {
+    if (!file(it).exists()) exit 1, "Missing file in TSV file: ${it}, see --help for more information"
+    return file(it)
+  }
+
+  // Return status [0,1]
+  // 0 == Normal, 1 == Tumor
+  static def returnStatus(it) {
+    if (!(it in [0, 1])) exit 1, "Status is not recognized in TSV file: ${it}, see --help for more information"
+    return it
+  }
+
+  // Return TSV if it has the correct number of items in row
+  static def returnTSV(it, number) {
+    if (it.size() != number) exit 1, "Malformed row in TSV file: ${it}, see --help for more information"
+    return it
+  }
+
+  // Sarek ascii art
+  static def sarek_ascii() {
+    println "    ____        _____                _     "
+    println "  .' _  `.     / ____|              | |    "
+    println " /  |\\`-_ \\   | (___   __ _ _ __ ___| | __ "
+    println "|   | \\  `-|   \\___ \\ / _` | '__/ __| |/ / "
+    println " \\ |   \\  /    ____) | (_| | | |  __|   <  "
+    println "  `|____\\'    |_____/ \\__,_|_|  \\___|_|\\_\\ "
+  }
+
 }