BiopetQScript.scala 6.41 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.utils.Logging
23
import org.broadinstitute.gatk.queue.{ QScript, QSettings }
Peter van 't Hof's avatar
Peter van 't Hof committed
24
import org.broadinstitute.gatk.queue.function.QFunction
Peter van 't Hof's avatar
Peter van 't Hof committed
25
import org.broadinstitute.gatk.queue.util.{ Logging => GatkLogging }
26

Peter van 't Hof's avatar
Peter van 't Hof committed
27
/** Base for biopet pipeline */
28
trait BiopetQScript extends Configurable with GatkLogging { qscript: QScript =>
29

30
  @Argument(doc = "JSON / YAML config file(s)", fullName = "config_file", shortName = "config", required = false)
31
  val configfiles: List[File] = Nil
bow's avatar
bow committed
32

Sander Bollen's avatar
Sander Bollen committed
33
  @Argument(doc = "Config values, value should be formatted like 'key=value' or 'namespace:namespace:key=value'", fullName = "config_value", shortName = "cv", required = false)
34 35
  val configValues: List[String] = Nil

Peter van 't Hof's avatar
Peter van 't Hof committed
36
  /** Output directory of pipeline */
Peter van 't Hof's avatar
Peter van 't Hof committed
37
  var outputDir: File = {
Peter van 't Hof's avatar
Peter van 't Hof committed
38 39
    if (config.contains("output_dir", path = Nil)) config("output_dir", path = Nil).asFile
    else new File(".")
40
  }
bow's avatar
bow committed
41

42
  @Argument(doc = "Disable all scatters", shortName = "DSC", required = false)
43
  var disableScatter: Boolean = false
44

bow's avatar
bow committed
45 46
  var outputFiles: Map[String, File] = Map()

Peter van 't Hof's avatar
Peter van 't Hof committed
47
  type InputFile = BiopetQScript.InputFile
48 49 50

  var inputFiles: List[InputFile] = Nil

Peter van 't Hof's avatar
Peter van 't Hof committed
51
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
Peter van 't Hof's avatar
Peter van 't Hof committed
52
  var qSettings: QSettings
bow's avatar
bow committed
53

Peter van 't Hof's avatar
Peter van 't Hof committed
54 55 56 57
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
  var functions: Seq[QFunction]

  /** Init for pipeline */
58
  def init()
bow's avatar
bow committed
59

Peter van 't Hof's avatar
Peter van 't Hof committed
60
  /** Pipeline itself */
61
  def biopetScript()
Peter van 't Hof's avatar
Peter van 't Hof committed
62

63 64 65
  /** Returns the extension to make the report */
  def reportClass: Option[ReportBuilderExtension] = None

Peter van 't Hof's avatar
Peter van 't Hof committed
66
  /** Script from queue itself, final to force some checks for each pipeline and write report */
67
  final def script() {
Peter van 't Hof's avatar
Peter van 't Hof committed
68 69
    outputDir = config("output_dir")
    outputDir = outputDir.getAbsoluteFile
70 71 72

    BiopetQScript.checkOutputDir(outputDir)

73 74
    init()
    biopetScript()
Peter van 't Hof's avatar
Peter van 't Hof committed
75
    logger.info("Biopet script done")
76

Peter van 't Hof's avatar
Peter van 't Hof committed
77 78 79 80 81 82
    if (disableScatter) {
      logger.info("Disable scatters")
      for (function <- functions) function match {
        case f: ScatterGatherableFunction => f.scatterCount = 1
        case _                            =>
      }
83
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
84 85 86 87

    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")
88 89 90 91
      case _ => reportClass.foreach { report =>
        report.mainFunction = true
        add(report)
      }
Peter van 't Hof's avatar
Peter van 't Hof committed
92 93
    }

Peter van 't Hof's avatar
Peter van 't Hof committed
94
    logger.info("Running pre commands")
95
    for (function <- functions) function match {
96
      case f: BiopetCommandLineFunction =>
97 98
        f.preProcessExecutable()
        f.beforeGraph()
Peter van 't Hof's avatar
Peter van 't Hof committed
99
        f.internalBeforeGraph()
Peter van 't Hof's avatar
Peter van 't Hof committed
100
        f.commandLine
Peter van 't Hof's avatar
Peter van 't Hof committed
101 102
      case f: WriteSummary => f.init()
      case _               =>
103
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
104

Peter van 't Hof's avatar
Peter van 't Hof committed
105 106
    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
107
    else Logging.addError("Parent of output dir: '" + outputDir.getParent + "' is not writeable, output directory cannot be created")
108

Peter van 't Hof's avatar
Peter van 't Hof committed
109 110
    logger.info("Checking input files")
    inputFiles.par.foreach { i =>
Peter van 't Hof's avatar
Peter van 't Hof committed
111
      if (!i.file.exists()) Logging.addError(s"Input file does not exist: ${i.file}")
112
      if (!i.file.canRead) Logging.addError(s"Input file can not be read: ${i.file}")
Sander Bollen's avatar
Sander Bollen committed
113
      if (!i.file.isAbsolute) Logging.addError(s"Input file should be an absolute path: ${i.file}")
114
    }
Peter van 't Hof's avatar
Peter van 't Hof committed
115

Peter van 't Hof's avatar
Peter van 't Hof committed
116
    functions.filter(_.jobOutputFile == null).foreach(f => {
Peter van 't Hof's avatar
Peter van 't Hof committed
117
      try {
Peter van 't Hof's avatar
Peter van 't Hof committed
118 119
        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
120 121 122
      } 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
123
    })
124

Peter van 't Hof's avatar
Peter van 't Hof committed
125
    WriteDependencies.writeDependencies(functions, new File(outputDir, ".log"), qSettings.runName)
126

Peter van 't Hof's avatar
Peter van 't Hof committed
127
    Logging.checkErrors()
Peter van 't Hof's avatar
Peter van 't Hof committed
128
    logger.info("Script complete without errors")
129
  }
bow's avatar
bow committed
130

Peter van 't Hof's avatar
Peter van 't Hof committed
131 132 133
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
  def add(functions: QFunction*)

134
  /** Get implemented from org.broadinstitute.gatk.queue.QScript */
Peter van 't Hof's avatar
Peter van 't Hof committed
135
  def addAll(functions: scala.Traversable[org.broadinstitute.gatk.queue.function.QFunction])
136

137
  /** Function to set isIntermediate and add in 1 line */
bow's avatar
bow committed
138
  def add(function: QFunction, isIntermediate: Boolean = false) {
139 140 141
    function.isIntermediate = isIntermediate
    add(function)
  }
142 143 144 145 146 147 148 149 150

  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
151
          case _                 =>
152
        }
Peter van 't Hof's avatar
Peter van 't Hof committed
153
      case that: BiopetQScript =>
154 155 156 157 158 159
        that.init()
        that.biopetScript()
      case _ => subPipeline.script
    }
    addAll(subPipeline.functions)
  }
160
}
Peter van 't Hof's avatar
Peter van 't Hof committed
161 162

object BiopetQScript {
Peter van 't Hof's avatar
Peter van 't Hof committed
163
  case class InputFile(file: File, md5: Option[String] = None)
164 165 166

  def checkOutputDir(outputDir: File): Unit = {
    // Sanity checks
Peter van 't Hof's avatar
Peter van 't Hof committed
167 168
    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}")
169
    outputDir.mkdir()
Peter van 't Hof's avatar
Peter van 't Hof committed
170 171
    require(outputDir.getAbsoluteFile.canRead, s"No premision to read outputdir: $outputDir")
    require(outputDir.getAbsoluteFile.canWrite, s"No premision to write outputdir: $outputDir")
172
  }
173
}