BiopetQScript.scala 6.62 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
 * license; For commercial users or users who do not want to follow the AGPL
 * license, please contact us to obtain a separate license.
 */
15 16 17
package nl.lumc.sasc.biopet.core

import java.io.File
18

Peter van 't Hof's avatar
Peter van 't Hof committed
19
import nl.lumc.sasc.biopet.core.summary.{SummaryQScript, WriteSummary}
Peter van 't Hof's avatar
Peter van 't Hof committed
20
import nl.lumc.sasc.biopet.utils.config.Configurable
21
import nl.lumc.sasc.biopet.core.report.ReportBuilderExtension
Peter van 't Hof's avatar
Peter van 't Hof committed
22
import nl.lumc.sasc.biopet.core.workaround.BiopetQCommandLine
Peter van 't Hof's avatar
Peter van 't Hof committed
23
import nl.lumc.sasc.biopet.utils.Logging
Peter van 't Hof's avatar
Peter van 't Hof committed
24
import org.broadinstitute.gatk.queue.{QScript, QSettings}
Peter van 't Hof's avatar
Peter van 't Hof committed
25
import org.broadinstitute.gatk.queue.function.QFunction
Peter van 't Hof's avatar
Peter van 't Hof committed
26
import org.broadinstitute.gatk.queue.util.{Logging => GatkLogging}
27

Peter van 't Hof's avatar
Peter van 't Hof committed
28
/** Base for biopet pipeline */
29
trait BiopetQScript extends Configurable with GatkLogging { qscript: QScript =>
30

31
  @Argument(doc = "JSON / YAML config file(s)", fullName = "config_file", shortName = "config", required = false)
32
  val configfiles: List[File] = Nil
bow's avatar
bow committed
33

Sander Bollen's avatar
Sander Bollen committed
34
  @Argument(doc = "Config values, value should be formatted like 'key=value' or 'namespace:namespace:key=value'", fullName = "config_value", shortName = "cv", required = false)
35 36
  val configValues: List[String] = Nil

Peter van 't Hof's avatar
Peter van 't Hof committed
37
  /** Output directory of pipeline */
Peter van 't Hof's avatar
Peter van 't Hof committed
38
  var outputDir: File = {
Peter van 't Hof's avatar
Peter van 't Hof committed
39 40
    if (config.contains("output_dir", path = Nil)) config("output_dir", path = Nil).asFile
    else new File(".")
41
  }
bow's avatar
bow committed
42

43
  @Argument(doc = "Disable all scatters", shortName = "DSC", required = false)
44
  var disableScatter: Boolean = false
45

bow's avatar
bow committed
46 47
  var outputFiles: Map[String, File] = Map()

Peter van 't Hof's avatar
Peter van 't Hof committed
48
  type InputFile = BiopetQScript.InputFile
49 50 51

  var inputFiles: List[InputFile] = Nil

Peter van 't Hof's avatar
Peter van 't Hof committed
52
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
Peter van 't Hof's avatar
Peter van 't Hof committed
53
  var qSettings: QSettings
bow's avatar
bow committed
54

Peter van 't Hof's avatar
Peter van 't Hof committed
55 56 57 58
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
  var functions: Seq[QFunction]

  /** Init for pipeline */
59
  def init()
bow's avatar
bow committed
60

Peter van 't Hof's avatar
Peter van 't Hof committed
61
  /** Pipeline itself */
62
  def biopetScript()
Peter van 't Hof's avatar
Peter van 't Hof committed
63

64 65 66
  /** Returns the extension to make the report */
  def reportClass: Option[ReportBuilderExtension] = None

67 68
  val skipWriteDependencies: Boolean = config("skip_write_dependencies", default = false)

Peter van 't Hof's avatar
Peter van 't Hof committed
69
  /** Script from queue itself, final to force some checks for each pipeline and write report */
70
  final def script() {
Peter van 't Hof's avatar
Peter van 't Hof committed
71 72
    outputDir = config("output_dir")
    outputDir = outputDir.getAbsoluteFile
73 74 75

    BiopetQScript.checkOutputDir(outputDir)

76 77
    init()
    biopetScript()
Peter van 't Hof's avatar
Peter van 't Hof committed
78
    logger.info("Biopet script done")
79

Peter van 't Hof's avatar
Peter van 't Hof committed
80 81 82 83 84 85
    if (disableScatter) {
      logger.info("Disable scatters")
      for (function <- functions) function match {
        case f: ScatterGatherableFunction => f.scatterCount = 1
        case _                            =>
      }
86
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
87 88 89 90

    this match {
      case q: MultiSampleQScript if q.onlySamples.nonEmpty && !q.samples.forall(x => q.onlySamples.contains(x._1)) =>
        logger.info("Write report is skipped because sample flag is used")
91 92 93
      case _ => reportClass.foreach { report =>
        add(report)
      }
Peter van 't Hof's avatar
Peter van 't Hof committed
94 95
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
96
    logger.info("Running pre commands")
97
    for (function <- functions) function match {
98
      case f: BiopetCommandLineFunction =>
99 100
        f.preProcessExecutable()
        f.beforeGraph()
Peter van 't Hof's avatar
Peter van 't Hof committed
101
        f.internalBeforeGraph()
Peter van 't Hof's avatar
Peter van 't Hof committed
102
        f.commandLine
Peter van 't Hof's avatar
Peter van 't Hof committed
103 104
      case f: WriteSummary => f.init()
      case _               =>
105
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
106

Peter van 't Hof's avatar
Peter van 't Hof committed
107 108
    if (outputDir.getParentFile.canWrite || (outputDir.exists && outputDir.canWrite))
      globalConfig.writeReport(qSettings.runName, new File(outputDir, ".log/" + qSettings.runName))
Sander Bollen's avatar
Sander Bollen committed
109
    else Logging.addError("Parent of output dir: '" + outputDir.getParent + "' is not writeable, output directory cannot be created")
110

Peter van 't Hof's avatar
Peter van 't Hof committed
111 112
    logger.info("Checking input files")
    inputFiles.par.foreach { i =>
Peter van 't Hof's avatar
Peter van 't Hof committed
113
      if (!i.file.exists()) Logging.addError(s"Input file does not exist: ${i.file}")
114
      if (!i.file.canRead) Logging.addError(s"Input file can not be read: ${i.file}")
Sander Bollen's avatar
Sander Bollen committed
115
      if (!i.file.isAbsolute) Logging.addError(s"Input file should be an absolute path: ${i.file}")
116
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
117

Peter van 't Hof's avatar
Peter van 't Hof committed
118
    functions.filter(_.jobOutputFile == null).foreach(f => {
Peter van 't Hof's avatar
Peter van 't Hof committed
119
      try {
Peter van 't Hof's avatar
Peter van 't Hof committed
120 121
        val className = if (f.getClass.isAnonymousClass) f.getClass.getSuperclass.getSimpleName else f.getClass.getSimpleName
        f.jobOutputFile = new File(f.firstOutput.getAbsoluteFile.getParent, "." + f.firstOutput.getName + "." + className + ".out")
Peter van 't Hof's avatar
Peter van 't Hof committed
122 123 124
      } catch {
        case e: NullPointerException => logger.warn(s"Can't generate a jobOutputFile for $f")
      }
Peter van 't Hof's avatar
Peter van 't Hof committed
125
    })
126

Peter van 't Hof's avatar
Peter van 't Hof committed
127 128 129
    if (!skipWriteDependencies) WriteDependencies.writeDependencies(
      functions,
      new File(outputDir, ".log" + File.separator + BiopetQCommandLine.timestamp + File.separator + "graph"))
130

Peter van 't Hof's avatar
Peter van 't Hof committed
131
    Logging.checkErrors()
Peter van 't Hof's avatar
Peter van 't Hof committed
132
    logger.info("Script complete without errors")
133
  }
bow's avatar
bow committed
134

Peter van 't Hof's avatar
Peter van 't Hof committed
135 136 137
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
  def add(functions: QFunction*)

138
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
Peter van 't Hof's avatar
Peter van 't Hof committed
139
  def addAll(functions: scala.Traversable[org.broadinstitute.gatk.queue.function.QFunction])
140

141
  /** Function to set isIntermediate and add in 1 line */
bow's avatar
bow committed
142
  def add(function: QFunction, isIntermediate: Boolean = false) {
143 144 145
    function.isIntermediate = isIntermediate
    add(function)
  }
146 147 148 149 150 151 152 153 154

  def add(subPipeline: QScript): Unit = {
    subPipeline.qSettings = this.qSettings
    subPipeline match {
      case that: SummaryQScript =>
        that.init()
        that.biopetScript()
        this match {
          case s: SummaryQScript => s.addSummaryQScript(that)
Peter van 't Hof's avatar
Peter van 't Hof committed
155
          case _                 =>
156
        }
Peter van 't Hof's avatar
Peter van 't Hof committed
157
      case that: BiopetQScript =>
158 159 160 161 162 163
        that.init()
        that.biopetScript()
      case _ => subPipeline.script
    }
    addAll(subPipeline.functions)
  }
164
}
Peter van 't Hof's avatar
Peter van 't Hof committed
165 166

object BiopetQScript {
Peter van 't Hof's avatar
Peter van 't Hof committed
167
  case class InputFile(file: File, md5: Option[String] = None)
168 169 170

  def checkOutputDir(outputDir: File): Unit = {
    // Sanity checks
Peter van 't Hof's avatar
Peter van 't Hof committed
171 172
    require(outputDir.getAbsoluteFile.getParentFile.canRead, s"No premision to read parent of outputdir: ${outputDir.getParentFile}")
    require(outputDir.getAbsoluteFile.getParentFile.canWrite, s"No premision to write parent of outputdir: ${outputDir.getParentFile}")
173
    outputDir.mkdir()
Peter van 't Hof's avatar
Peter van 't Hof committed
174 175
    require(outputDir.getAbsoluteFile.canRead, s"No premision to read outputdir: $outputDir")
    require(outputDir.getAbsoluteFile.canWrite, s"No premision to write outputdir: $outputDir")
176
  }
Peter van 't Hof's avatar
Peter van 't Hof committed
177 178


179
}