* added optional 'no validation' for the Data Processing pipeline.
* some simplifications on the picard classes
This commit is contained in:
Mauricio Carneiro
parent
4cf7a2af23
commit
2b465ab43b
|
|
@ -72,6 +72,9 @@ class DataProcessingPipeline extends QScript {
|
|||
@Input(doc="Number of threads BWA should use", fullName="bwa_threads", shortName="bt", required=false)
|
||||
var bwaThreads: Int = 1
|
||||
|
||||
@Input(doc="Dont perform validation on the BAM files", fullName="no_validation", shortName="nv", required=false)
|
||||
var noValidation: Boolean = false
|
||||
|
||||
|
||||
/****************************************************************************
|
||||
* Global Variables
|
||||
|
|
@ -242,7 +245,12 @@ class DataProcessingPipeline extends QScript {
|
|||
val preValidateLog = swapExt(bam, ".bam", ".pre.validation")
|
||||
val postValidateLog = swapExt(bam, ".bam", ".post.validation")
|
||||
|
||||
add(validate(bam, preValidateLog))
|
||||
// Validation is an optional step for the BAM file generated after
|
||||
// alignment and the final bam file of the pipeline.
|
||||
if (!noValidation) {
|
||||
add(validate(bam, preValidateLog),
|
||||
validate(recalBam, postValidateLog))
|
||||
}
|
||||
|
||||
if (cleaningModel != ConsensusDeterminationModel.KNOWNS_ONLY)
|
||||
add(target(bam, targetIntervals))
|
||||
|
|
@ -253,8 +261,8 @@ class DataProcessingPipeline extends QScript {
|
|||
recal(dedupedBam, preRecalFile, recalBam),
|
||||
cov(recalBam, postRecalFile),
|
||||
analyzeCovariates(preRecalFile, preOutPath),
|
||||
analyzeCovariates(postRecalFile, postOutPath),
|
||||
validate(recalBam, postValidateLog))
|
||||
analyzeCovariates(postRecalFile, postOutPath))
|
||||
|
||||
|
||||
cohortList :+= recalBam
|
||||
}
|
||||
|
|
@ -278,6 +286,13 @@ class DataProcessingPipeline extends QScript {
|
|||
this.isIntermediate = true
|
||||
}
|
||||
|
||||
// General arguments to non-GATK tools
|
||||
trait ExternalCommonArgs extends CommandLineFunction {
|
||||
this.memoryLimit = 4
|
||||
this.isIntermediate = true
|
||||
}
|
||||
|
||||
|
||||
case class target (inBams: File, outIntervals: File) extends RealignerTargetCreator with CommandLineGATKArgs {
|
||||
if (cleaningModel != ConsensusDeterminationModel.KNOWNS_ONLY)
|
||||
this.input_file :+= inBams
|
||||
|
|
@ -296,8 +311,8 @@ class DataProcessingPipeline extends QScript {
|
|||
this.targetIntervals = tIntervals
|
||||
this.out = outBam
|
||||
this.rodBind :+= RodBind("dbsnp", "VCF", dbSNP)
|
||||
if (!indels.isEmpty)
|
||||
this.rodBind :+= RodBind("indels", "VCF", indels)
|
||||
if (!qscript.indels.isEmpty)
|
||||
this.rodBind :+= RodBind("indels", "VCF", qscript.indels)
|
||||
this.consensusDeterminationModel = consensusDeterminationModel
|
||||
this.compress = 0
|
||||
this.scatterCount = nContigs
|
||||
|
|
@ -328,7 +343,6 @@ class DataProcessingPipeline extends QScript {
|
|||
this.isIntermediate = false
|
||||
this.analysisName = queueLogDir + outBam + ".recalibration"
|
||||
this.jobName = queueLogDir + outBam + ".recalibration"
|
||||
|
||||
}
|
||||
|
||||
|
||||
|
|
@ -346,48 +360,41 @@ class DataProcessingPipeline extends QScript {
|
|||
this.jobName = queueLogDir + inRecalFile + ".analyze_covariates"
|
||||
}
|
||||
|
||||
case class dedup (inBam: File, outBam: File, metricsFile: File) extends MarkDuplicates {
|
||||
case class dedup (inBam: File, outBam: File, metricsFile: File) extends MarkDuplicates with ExternalCommonArgs {
|
||||
this.input = List(inBam)
|
||||
this.output = outBam
|
||||
this.metrics = metricsFile
|
||||
this.memoryLimit = 6
|
||||
this.isIntermediate = true
|
||||
this.analysisName = queueLogDir + outBam + ".dedup"
|
||||
this.jobName = queueLogDir + outBam + ".dedup"
|
||||
}
|
||||
|
||||
case class joinBams (inBams: List[File], outBam: File) extends MergeSamFiles {
|
||||
case class joinBams (inBams: List[File], outBam: File) extends MergeSamFiles with ExternalCommonArgs {
|
||||
this.input = inBams
|
||||
this.output = outBam
|
||||
this.memoryLimit = 4
|
||||
this.isIntermediate = true
|
||||
this.analysisName = queueLogDir + outBam + ".joinBams"
|
||||
this.jobName = queueLogDir + outBam + ".joinBams"
|
||||
}
|
||||
|
||||
case class sortSam (inSam: File, outBam: File, sortOrderP: SortOrder) extends SortSam {
|
||||
case class sortSam (inSam: File, outBam: File, sortOrderP: SortOrder) extends SortSam with ExternalCommonArgs {
|
||||
this.input = List(inSam)
|
||||
this.output = outBam
|
||||
this.sortOrder = sortOrderP
|
||||
this.memoryLimit = 4
|
||||
this.isIntermediate = true
|
||||
this.analysisName = queueLogDir + outBam + ".sortSam"
|
||||
this.jobName = queueLogDir + outBam + ".sortSam"
|
||||
}
|
||||
|
||||
case class validate (inBam: File, outLog: File) extends ValidateSamFile {
|
||||
case class validate (inBam: File, outLog: File) extends ValidateSamFile with ExternalCommonArgs {
|
||||
this.input = List(inBam)
|
||||
this.output = outLog
|
||||
this.maxRecordsInRam = 100000
|
||||
this.REFERENCE_SEQUENCE = qscript.reference
|
||||
this.memoryLimit = 4
|
||||
this.isIntermediate = false
|
||||
this.analysisName = queueLogDir + outLog + ".validate"
|
||||
this.jobName = queueLogDir + outLog + ".validate"
|
||||
}
|
||||
|
||||
|
||||
case class addReadGroup (inBam: File, outBam: File, readGroup: ReadGroup) extends AddOrReplaceReadGroups {
|
||||
case class addReadGroup (inBam: File, outBam: File, readGroup: ReadGroup) extends AddOrReplaceReadGroups with ExternalCommonArgs {
|
||||
this.input = List(inBam)
|
||||
this.output = outBam
|
||||
this.RGID = readGroup.id
|
||||
|
|
@ -403,12 +410,7 @@ class DataProcessingPipeline extends QScript {
|
|||
this.jobName = queueLogDir + outBam + ".rg"
|
||||
}
|
||||
|
||||
trait BWACommonArgs extends CommandLineFunction {
|
||||
this.memoryLimit = 4
|
||||
this.isIntermediate = true
|
||||
}
|
||||
|
||||
case class bwa_aln_se (inBam: File, outSai: File) extends CommandLineFunction with BWACommonArgs {
|
||||
case class bwa_aln_se (inBam: File, outSai: File) extends CommandLineFunction with ExternalCommonArgs {
|
||||
@Input(doc="bam file to be aligned") var bam = inBam
|
||||
@Output(doc="output sai file") var sai = outSai
|
||||
def commandLine = bwaPath + " aln -t " + bwaThreads + " -q 5 " + reference + " -b " + bam + " > " + sai
|
||||
|
|
@ -416,7 +418,7 @@ class DataProcessingPipeline extends QScript {
|
|||
this.jobName = queueLogDir + outSai + ".bwa_aln_se"
|
||||
}
|
||||
|
||||
case class bwa_aln_pe (inBam: File, outSai1: File, index: Int) extends CommandLineFunction with BWACommonArgs {
|
||||
case class bwa_aln_pe (inBam: File, outSai1: File, index: Int) extends CommandLineFunction with ExternalCommonArgs {
|
||||
@Input(doc="bam file to be aligned") var bam = inBam
|
||||
@Output(doc="output sai file for 1st mating pair") var sai = outSai1
|
||||
def commandLine = bwaPath + " aln -t " + bwaThreads + " -q 5 " + reference + " -b" + index + " " + bam + " > " + sai
|
||||
|
|
@ -424,7 +426,7 @@ class DataProcessingPipeline extends QScript {
|
|||
this.jobName = queueLogDir + outSai1 + ".bwa_aln_pe1"
|
||||
}
|
||||
|
||||
case class bwa_sam_se (inBam: File, inSai: File, outBam: File) extends CommandLineFunction with BWACommonArgs {
|
||||
case class bwa_sam_se (inBam: File, inSai: File, outBam: File) extends CommandLineFunction with ExternalCommonArgs {
|
||||
@Input(doc="bam file to be aligned") var bam = inBam
|
||||
@Input(doc="bwa alignment index file") var sai = inSai
|
||||
@Output(doc="output aligned bam file") var alignedBam = outBam
|
||||
|
|
@ -433,7 +435,7 @@ class DataProcessingPipeline extends QScript {
|
|||
this.jobName = queueLogDir + outBam + ".bwa_sam_se"
|
||||
}
|
||||
|
||||
case class bwa_sam_pe (inBam: File, inSai1: File, inSai2:File, outBam: File) extends CommandLineFunction with BWACommonArgs {
|
||||
case class bwa_sam_pe (inBam: File, inSai1: File, inSai2:File, outBam: File) extends CommandLineFunction with ExternalCommonArgs {
|
||||
@Input(doc="bam file to be aligned") var bam = inBam
|
||||
@Input(doc="bwa alignment index file for 1st mating pair") var sai1 = inSai1
|
||||
@Input(doc="bwa alignment index file for 2nd mating pair") var sai2 = inSai2
|
||||
|
|
|
|||
Loading…
Reference in New Issue