GearsSingle.scala 6.36 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
/**
 * 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
 *
11
 * A dual licensing mode is applied. The source code within this project is freely available for non-commercial use under an AGPL
12
13
14
15
16
 * 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.pipelines.gears

Peter van 't Hof's avatar
Peter van 't Hof committed
17
import nl.lumc.sasc.biopet.core.summary.SummaryQScript
Peter van 't Hof's avatar
Peter van 't Hof committed
18
import nl.lumc.sasc.biopet.core.BiopetQScript.InputFile
Peter van 't Hof's avatar
Peter van 't Hof committed
19
import nl.lumc.sasc.biopet.core.{ PipelineCommand, SampleLibraryTag }
Peter van 't Hof's avatar
Peter van 't Hof committed
20
import nl.lumc.sasc.biopet.pipelines.flexiprep.Flexiprep
Peter van 't Hof's avatar
Peter van 't Hof committed
21
import nl.lumc.sasc.biopet.utils.Logging
Peter van 't Hof's avatar
Peter van 't Hof committed
22
import nl.lumc.sasc.biopet.utils.config.Configurable
23
24
import org.broadinstitute.gatk.queue.QScript

25
/**
Wai Yi Leung's avatar
Wai Yi Leung committed
26
 * Created by wyleung
27
 */
28
class GearsSingle(val root: Configurable) extends QScript with SummaryQScript with SampleLibraryTag {
29
  def this() = this(null)
30

Wai Yi Leung's avatar
Wai Yi Leung committed
31
32
  @Input(doc = "R1 reads in FastQ format", shortName = "R1", required = false)
  var fastqR1: Option[File] = None
33

Wai Yi Leung's avatar
Wai Yi Leung committed
34
35
  @Input(doc = "R2 reads in FastQ format", shortName = "R2", required = false)
  var fastqR2: Option[File] = None
36

Wai Yi Leung's avatar
Wai Yi Leung committed
37
  @Input(doc = "All unmapped reads will be extracted from this bam for analysis", shortName = "bam", required = false)
Peter van 't Hof's avatar
Peter van 't Hof committed
38
  var bamFile: Option[File] = None
39

Peter van 't Hof's avatar
Peter van 't Hof committed
40
41
  @Argument(required = false)
  var outputName: String = _
42

Peter van 't Hof's avatar
Peter van 't Hof committed
43
  lazy val krakenScript = if (config("gears_use_kraken", default = true)) Some(new GearsKraken(this)) else None
Peter van 't Hof's avatar
Peter van 't Hof committed
44
45
46
  lazy val qiimeRatx = if (config("gears_use_qiime_rtax", default = false)) Some(new GearsQiimeRtax(this)) else None
  lazy val qiimeClosed = if (config("gears_use_qiime_closed", default = false)) Some(new GearsQiimeClosed(this)) else None
  lazy val seqCount = if (config("gears_use_seq_count", default = false)) Some(new GearsSeqCount(this)) else None
Peter van 't Hof's avatar
Peter van 't Hof committed
47

Peter van 't Hof's avatar
Peter van 't Hof committed
48
49
  /** Executed before running the script */
  def init(): Unit = {
50
    require(fastqR1.isDefined || bamFile.isDefined, "Please specify fastq-file(s) or bam file")
Peter van 't Hof's avatar
Peter van 't Hof committed
51
    require(fastqR1.isDefined != bamFile.isDefined, "Provide either a bam file or a R1/R2 file")
Peter van 't Hof's avatar
Peter van 't Hof committed
52
    if (sampleId == null || sampleId == None) Logging.addError("Missing sample ID on GearsSingle module")
Peter van 't Hof's avatar
Peter van 't Hof committed
53
54

    if (outputName == null) {
Wai Yi Leung's avatar
Wai Yi Leung committed
55
      if (fastqR1.isDefined) outputName = fastqR1.map(_.getName
Peter van 't Hof's avatar
Peter van 't Hof committed
56
57
58
59
60
        .stripSuffix(".gz")
        .stripSuffix(".fastq")
        .stripSuffix(".fq"))
        .getOrElse("noName")
      else outputName = bamFile.map(_.getName.stripSuffix(".bam")).getOrElse("noName")
61
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
62
63
64
65

    if (fastqR1.isDefined) {
      fastqR1.foreach(inputFiles :+= InputFile(_))
      fastqR2.foreach(inputFiles :+= InputFile(_))
Peter van 't Hof's avatar
Peter van 't Hof committed
66
    } else inputFiles :+= InputFile(bamFile.get)
Peter van 't Hof's avatar
Peter van 't Hof committed
67
  }
68

Wai Yi Leung's avatar
Wai Yi Leung committed
69
  override def reportClass = {
70
    val gears = new GearsSingleReport(this)
Wai Yi Leung's avatar
Wai Yi Leung committed
71
72
    gears.outputDir = new File(outputDir, "report")
    gears.summaryFile = summaryFile
73
74
    sampleId.foreach(gears.args += "sampleId" -> _)
    libId.foreach(gears.args += "libId" -> _)
Wai Yi Leung's avatar
Wai Yi Leung committed
75
76
    Some(gears)
  }
77

Peter van 't Hof's avatar
Peter van 't Hof committed
78
79
  protected var skipFlexiprep: Boolean = config("skip_flexiprep", default = false)

80
  protected def executeFlexiprep(r1: File, r2: Option[File]): (File, Option[File]) = {
Peter van 't Hof's avatar
Peter van 't Hof committed
81
    if (!skipFlexiprep) {
82
      val flexiprep = new Flexiprep(this)
Sander van der Zeeuw's avatar
Sander van der Zeeuw committed
83
      flexiprep.inputR1 = r1
84
      flexiprep.inputR2 = r2
Peter van 't Hof's avatar
Peter van 't Hof committed
85
86
      flexiprep.sampleId = if (sampleId.isEmpty) Some("noSampleName") else sampleId
      flexiprep.libId = if (libId.isEmpty) Some("noLibName") else libId
87
88
89
90
91
92
      flexiprep.outputDir = new File(outputDir, "flexiprep")
      add(flexiprep)
      (flexiprep.fastqR1Qc, flexiprep.fastqR2Qc)
    } else (r1, r2)
  }

Peter van 't Hof's avatar
Peter van 't Hof committed
93
94
  /** Method to add jobs */
  def biopetScript(): Unit = {
Peter van 't Hof's avatar
Peter van 't Hof committed
95
    val (r1, r2): (File, Option[File]) = (fastqR1, fastqR2, bamFile) match {
96
      case (Some(r1), _, _) => executeFlexiprep(r1, fastqR2)
Peter van 't Hof's avatar
Peter van 't Hof committed
97
98
      case (_, _, Some(bam)) =>
        val extract = new ExtractUnmappedReads(this)
Peter van 't Hof's avatar
Peter van 't Hof committed
99
        extract.outputDir = outputDir
Peter van 't Hof's avatar
Peter van 't Hof committed
100
        extract.bamFile = bam
Peter van 't Hof's avatar
Peter van 't Hof committed
101
        extract.outputName = outputName
Peter van 't Hof's avatar
Peter van 't Hof committed
102
        add(extract)
Peter van 't Hof's avatar
Peter van 't Hof committed
103
        executeFlexiprep(extract.fastqUnmappedR1, extract.fastqUnmappedR2)
Peter van 't Hof's avatar
Peter van 't Hof committed
104
      case _ => throw new IllegalArgumentException("Missing input files")
Peter van 't Hof's avatar
Peter van 't Hof committed
105
106
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
107
108
109
    lazy val combinedFastq = {
      r2 match {
        case Some(r2) =>
110
111
112
113
114
115
          val combineReads = new CombineReads(this)
          combineReads.outputDir = new File(outputDir, "combine_reads")
          combineReads.fastqR1 = r1
          combineReads.fastqR2 = r2
          add(combineReads)
          combineReads.combinedFastq
Peter van 't Hof's avatar
Peter van 't Hof committed
116
117
118
119
        case _ => r1
      }
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
120
    krakenScript foreach { kraken =>
Peter van 't Hof's avatar
Peter van 't Hof committed
121
      kraken.outputDir = new File(outputDir, "kraken")
122
123
      kraken.fastqR1 = r1
      kraken.fastqR2 = r2
Peter van 't Hof's avatar
Peter van 't Hof committed
124
      kraken.outputName = outputName
Peter van 't Hof's avatar
Peter van 't Hof committed
125
      add(kraken)
Peter van 't Hof's avatar
Peter van 't Hof committed
126
127
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
128
    qiimeRatx foreach { qiimeRatx =>
Peter van 't Hof's avatar
Peter van 't Hof committed
129
      qiimeRatx.outputDir = new File(outputDir, "qiime_rtax")
130
131
      qiimeRatx.fastqR1 = r1
      qiimeRatx.fastqR2 = r2
Peter van 't Hof's avatar
Peter van 't Hof committed
132
      add(qiimeRatx)
Peter van 't Hof's avatar
Peter van 't Hof committed
133
    }
134

Peter van 't Hof's avatar
Peter van 't Hof committed
135
    qiimeClosed foreach { qiimeClosed =>
Peter van 't Hof's avatar
Peter van 't Hof committed
136
      qiimeClosed.outputDir = new File(outputDir, "qiime_closed")
Peter van 't Hof's avatar
Peter van 't Hof committed
137
      qiimeClosed.fastqInput = combinedFastq
Peter van 't Hof's avatar
Peter van 't Hof committed
138
      add(qiimeClosed)
139
140
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
141
142
143
144
145
146
    seqCount.foreach { seqCount =>
      seqCount.fastqInput = combinedFastq
      seqCount.outputDir = new File(outputDir, "seq_count")
      add(seqCount)
    }

Wai Yi Leung's avatar
Wai Yi Leung committed
147
    addSummaryJobs()
148
  }
Peter van 't Hof's avatar
Peter van 't Hof committed
149
150

  /** Location of summary file */
Peter van 't Hof's avatar
Peter van 't Hof committed
151
  def summaryFile = new File(outputDir, sampleId.getOrElse("sampleName_unknown") + ".gears.summary.json")
Peter van 't Hof's avatar
Peter van 't Hof committed
152

153
  /** Pipeline settings shown in the summary file */
Peter van 't Hof's avatar
Peter van 't Hof committed
154
155
156
157
158
159
  def summarySettings: Map[String, Any] = Map(
    "skip_flexiprep" -> skipFlexiprep,
    "gears_use_kraken" -> krakenScript.isDefined,
    "gear_use_qiime_rtax" -> qiimeRatx.isDefined,
    "gear_use_qiime_closed" -> qiimeClosed.isDefined
  )
Peter van 't Hof's avatar
Peter van 't Hof committed
160

161
  /** Statistics shown in the summary file */
Wai Yi Leung's avatar
Wai Yi Leung committed
162
163
  def summaryFiles: Map[String, File] = Map.empty ++
    (if (bamFile.isDefined) Map("input_bam" -> bamFile.get) else Map()) ++
Wai Yi Leung's avatar
Wai Yi Leung committed
164
    (if (fastqR1.isDefined) Map("input_R1" -> fastqR1.get) else Map()) ++
Peter van 't Hof's avatar
Peter van 't Hof committed
165
    (if (fastqR2.isDefined) Map("input_R2" -> fastqR2.get) else Map()) ++
Wai Yi Leung's avatar
Wai Yi Leung committed
166
    outputFiles
167
168
169
}

/** This object give a default main method to the pipelines */
170
object GearsSingle extends PipelineCommand