Skip to content
Snippets Groups Projects
Unverified Commit ca35e74a authored by Cats's avatar Cats Committed by GitHub
Browse files

Merge pull request #92 from biowdl/BIOWDL-199 

Dockerize GATK. Scattering now works with containers. Tabix another docker image.
parents e55fc372 68ff64af
No related branches found
No related tags found
No related merge requests found
Hide whitespace changes
Inline Side-by-side
biopet/biopet.wdl
+ 20
− 13
View file @ ca35e74a
...@@ -205,54 +205,60 @@ task FastqSync { ...@@ -205,54 +205,60 @@ task FastqSync {
task ReorderGlobbedScatters { task ReorderGlobbedScatters {
input { input {
Array[File]+ scatters Array[File]+ scatters
String scatterDir # Should not be changed from the main pipeline. As it should not influence results.
String dockerTag = "3.6"
} }
command <<< command <<<
set -e
# Copy all the scatter files to the CWD so the output matches paths in
# the cwd.
for file in ~{sep=" " scatters}
do cp $file .
done
python << CODE python << CODE
from os.path import basename from os.path import basename
scatters = ['~{sep="','" scatters}'] scatters = ['~{sep="','" scatters}']
splitext = [basename(x).split(".") for x in scatters] splitext = [basename(x).split(".") for x in scatters]
splitnum = [x.split("-") + [y] for x,y in splitext] splitnum = [x.split("-") + [y] for x,y in splitext]
ordered = sorted(splitnum, key=lambda x: int(x[1])) ordered = sorted(splitnum, key=lambda x: int(x[1]))
merged = ["~{scatterDir}/{}-{}.{}".format(x[0],x[1],x[2]) for x in ordered] merged = ["{}-{}.{}".format(x[0],x[1],x[2]) for x in ordered]
for x in merged: for x in merged:
print(x) print(x)
CODE CODE
>>> >>>
output { output {
Array[String] reorderedScatters = read_lines(stdout()) Array[File] reorderedScatters = read_lines(stdout())
} }
runtime { runtime {
memory: 1 docker: "python:" + dockerTag
# 4 gigs of memory to be able to build the docker image in singularity
memory: 4
} }
} }
task ScatterRegions { task ScatterRegions {
input { input {
String? preCommand
Reference reference Reference reference
String outputDirPath
File? toolJar
Int? scatterSize Int? scatterSize
File? regions File? regions
Boolean notSplitContigs = false Boolean notSplitContigs = false
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3.0 Float memoryMultiplier = 3.0
String dockerTag = "0.2--0"
} }
String toolCommand = if defined(toolJar) # OutDirPath must be defined here because the glob process relies on
then "java -Xmx" + memory + "G -jar " +toolJar # linking. This path must be in the containers filesystem, otherwise the
else "biopet-scatterregions -Xmx" + memory + "G" # linking does not work.
String outputDirPath = "scatters"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand}
mkdir -p ~{outputDirPath} mkdir -p ~{outputDirPath}
~{toolCommand} \ biopet-scatterregions -Xmx~{memory}G \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-o ~{outputDirPath} \ -o ~{outputDirPath} \
~{"-s " + scatterSize} \ ~{"-s " + scatterSize} \
...@@ -265,6 +271,7 @@ task ScatterRegions { ...@@ -265,6 +271,7 @@ task ScatterRegions {
} }
runtime { runtime {
docker: "quay.io/biocontainers/biopet-scatterregions:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
......
gatk.wdl
+ 45
− 87
View file @ ca35e74a
...@@ -5,8 +5,6 @@ import "common.wdl" ...@@ -5,8 +5,6 @@ import "common.wdl"
# Apply Base Quality Score Recalibration (BQSR) model # Apply Base Quality Score Recalibration (BQSR) model
task ApplyBQSR { task ApplyBQSR {
input { input {
String? preCommand
File? gatkJar
IndexedBamFile inputBam IndexedBamFile inputBam
String outputBamPath String outputBamPath
File recalibrationReport File recalibrationReport
...@@ -15,28 +13,25 @@ task ApplyBQSR { ...@@ -15,28 +13,25 @@ task ApplyBQSR {
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3.0 Float memoryMultiplier = 3.0
String dockerTag = "4.1.0.0--0"
} }
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputBamPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
ApplyBQSR \ ApplyBQSR \
--create-output-bam-md5 \ --create-output-bam-md5 \
--add-output-sam-program-record \ --add-output-sam-program-record \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-I ~{inputBam.file} \ -I ~{inputBam.file} \
--use-original-qualities \ --use-original-qualities \
-O ~{outputBamPath} \ -O ~{outputBamPath} \
-bqsr ~{recalibrationReport} \ -bqsr ~{recalibrationReport} \
--static-quantized-quals 10 \ --static-quantized-quals 10 \
--static-quantized-quals 20 \ --static-quantized-quals 20 \
--static-quantized-quals 30 \ --static-quantized-quals 30 \
-L ~{sep=" -L " sequenceGroupInterval} -L ~{sep=" -L " sequenceGroupInterval}
} }
output { output {
...@@ -48,6 +43,7 @@ task ApplyBQSR { ...@@ -48,6 +43,7 @@ task ApplyBQSR {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
...@@ -55,8 +51,6 @@ task ApplyBQSR { ...@@ -55,8 +51,6 @@ task ApplyBQSR {
# Generate Base Quality Score Recalibration (BQSR) model # Generate Base Quality Score Recalibration (BQSR) model
task BaseRecalibrator { task BaseRecalibrator {
input { input {
String? preCommand
File? gatkJar
IndexedBamFile inputBam IndexedBamFile inputBam
String recalibrationReportPath String recalibrationReportPath
Array[File]+ sequenceGroupInterval Array[File]+ sequenceGroupInterval
...@@ -66,6 +60,7 @@ task BaseRecalibrator { ...@@ -66,6 +60,7 @@ task BaseRecalibrator {
Reference reference Reference reference
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3.0 Float memoryMultiplier = 3.0
String dockerTag = "4.1.0.0--0"
} }
Array[File]+ knownIndelsSitesVCFsArg = flatten([ Array[File]+ knownIndelsSitesVCFsArg = flatten([
...@@ -73,14 +68,10 @@ task BaseRecalibrator { ...@@ -73,14 +68,10 @@ task BaseRecalibrator {
[select_first([dbsnpVCF]).file] [select_first([dbsnpVCF]).file]
]) ])
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{recalibrationReportPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
BaseRecalibrator \ BaseRecalibrator \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-I ~{inputBam.file} \ -I ~{inputBam.file} \
...@@ -95,35 +86,28 @@ task BaseRecalibrator { ...@@ -95,35 +86,28 @@ task BaseRecalibrator {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
task CombineGVCFs { task CombineGVCFs {
input { input {
String? preCommand
Array[File]+ gvcfFiles Array[File]+ gvcfFiles
Array[File]+ gvcfFilesIndex Array[File]+ gvcfFilesIndex
Array[File]+ intervals Array[File]+ intervals
String outputPath String outputPath
String? gatkJar
Reference reference Reference reference
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3.0 Float memoryMultiplier = 3.0
String dockerTag = "4.1.0.0--0"
} }
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
CombineGVCFs \ CombineGVCFs \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-O ~{outputPath} \ -O ~{outputPath} \
...@@ -139,6 +123,7 @@ task CombineGVCFs { ...@@ -139,6 +123,7 @@ task CombineGVCFs {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
...@@ -146,23 +131,18 @@ task CombineGVCFs { ...@@ -146,23 +131,18 @@ task CombineGVCFs {
# Combine multiple recalibration tables from scattered BaseRecalibrator runs # Combine multiple recalibration tables from scattered BaseRecalibrator runs
task GatherBqsrReports { task GatherBqsrReports {
input { input {
String? preCommand
String? gatkJar
Array[File] inputBQSRreports Array[File] inputBQSRreports
String outputReportPath String outputReportPath
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3.0 Float memoryMultiplier = 3.0
String dockerTag = "4.1.0.0--0"
} }
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputReportPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
GatherBQSRReports \ GatherBQSRReports \
-I ~{sep=' -I ' inputBQSRreports} \ -I ~{sep=' -I ' inputBQSRreports} \
-O ~{outputReportPath} -O ~{outputReportPath}
...@@ -173,39 +153,31 @@ task GatherBqsrReports { ...@@ -173,39 +153,31 @@ task GatherBqsrReports {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
task GenotypeGVCFs { task GenotypeGVCFs {
input { input {
String? preCommand
Array[File]+ gvcfFiles Array[File]+ gvcfFiles
Array[File]+ gvcfFilesIndex Array[File]+ gvcfFilesIndex
Array[File]+ intervals Array[File]+ intervals
String outputPath String outputPath
String? gatkJar
Reference reference Reference reference
IndexedVcfFile? dbsnpVCF IndexedVcfFile? dbsnpVCF
Int memory = 6 Int memory = 6
Float memoryMultiplier = 2.0 Float memoryMultiplier = 2.0
String dockerTag = "4.1.0.0--0"
} }
File dbsnpFile = if (defined(dbsnpVCF)) then select_first([dbsnpVCF]).file else "" File dbsnpFile = if (defined(dbsnpVCF)) then select_first([dbsnpVCF]).file else ""
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
GenotypeGVCFs \ GenotypeGVCFs \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-O ~{outputPath} \ -O ~{outputPath} \
...@@ -224,7 +196,8 @@ task GenotypeGVCFs { ...@@ -224,7 +196,8 @@ task GenotypeGVCFs {
} }
} }
runtime{ runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
...@@ -232,31 +205,25 @@ task GenotypeGVCFs { ...@@ -232,31 +205,25 @@ task GenotypeGVCFs {
# Call variants on a single sample with HaplotypeCaller to produce a GVCF # Call variants on a single sample with HaplotypeCaller to produce a GVCF
task HaplotypeCallerGvcf { task HaplotypeCallerGvcf {
input { input {
String? preCommand
Array[File]+ inputBams Array[File]+ inputBams
Array[File]+ inputBamsIndex Array[File]+ inputBamsIndex
Array[File]+ intervalList Array[File]+ intervalList
String gvcfPath String gvcfPath
Reference reference Reference reference
Float contamination = 0.0 Float contamination = 0.0
String? gatkJar
IndexedVcfFile? dbsnpVCF IndexedVcfFile? dbsnpVCF
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3 Float memoryMultiplier = 3
String dockerTag = "4.1.0.0--0"
} }
File dbsnpFile = if (defined(dbsnpVCF)) then select_first([dbsnpVCF]).file else "" File dbsnpFile = if (defined(dbsnpVCF)) then select_first([dbsnpVCF]).file else ""
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{gvcfPath})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
HaplotypeCaller \ HaplotypeCaller \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-O ~{gvcfPath} \ -O ~{gvcfPath} \
...@@ -275,14 +242,13 @@ task HaplotypeCallerGvcf { ...@@ -275,14 +242,13 @@ task HaplotypeCallerGvcf {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
task MuTect2 { task MuTect2 {
input { input {
String? preCommand
Array[File]+ inputBams Array[File]+ inputBams
Array[File]+ inputBamsIndex Array[File]+ inputBamsIndex
Reference reference Reference reference
...@@ -291,19 +257,15 @@ task MuTect2 { ...@@ -291,19 +257,15 @@ task MuTect2 {
String? normalSample String? normalSample
Array[File]+ intervals Array[File]+ intervals
String? gatkJar
Int memory = 4 Int memory = 4
Float memoryMultiplier = 3 Float memoryMultiplier = 3
String dockerTag = "4.1.0.0--0"
} }
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputVcf})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
Mutect2 \ Mutect2 \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
-I ~{sep=" -I " inputBams} \ -I ~{sep=" -I " inputBams} \
...@@ -321,32 +283,27 @@ task MuTect2 { ...@@ -321,32 +283,27 @@ task MuTect2 {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
task SplitNCigarReads { task SplitNCigarReads {
input { input {
String? preCommand
IndexedBamFile inputBam IndexedBamFile inputBam
Reference reference Reference reference
String outputBam String outputBam
String? gatkJar
Array[File]+ intervals Array[File]+ intervals
Int memory = 4 Int memory = 4
Float memoryMultiplier = 4 Float memoryMultiplier = 4
String dockerTag = "4.1.0.0--0"
} }
String toolCommand = if defined(gatkJar)
then "java -Xmx" + memory + "G -jar " + gatkJar
else "gatk --java-options -Xmx" + memory + "G"
command { command {
set -e -o pipefail set -e -o pipefail
~{preCommand} mkdir -p $(dirname ~{outputBam})
~{toolCommand} \ gatk --java-options -Xmx~{memory}G \
SplitNCigarReads \ SplitNCigarReads \
-I ~{inputBam.file} \ -I ~{inputBam.file} \
-R ~{reference.fasta} \ -R ~{reference.fasta} \
...@@ -362,6 +319,7 @@ task SplitNCigarReads { ...@@ -362,6 +319,7 @@ task SplitNCigarReads {
} }
runtime { runtime {
docker: "quay.io/biocontainers/gatk4:" + dockerTag
memory: ceil(memory * memoryMultiplier) memory: ceil(memory * memoryMultiplier)
} }
} }
samtools.wdl
+ 10
− 0
View file @ ca35e74a
...@@ -7,6 +7,7 @@ task BgzipAndIndex { ...@@ -7,6 +7,7 @@ task BgzipAndIndex {
File inputFile File inputFile
String outputDir String outputDir
String type = "vcf" String type = "vcf"
String dockerTag = "0.2.6--ha92aebf_0"
} }
String outputGz = outputDir + "/" + basename(inputFile) + ".gz" String outputGz = outputDir + "/" + basename(inputFile) + ".gz"
...@@ -20,6 +21,10 @@ task BgzipAndIndex { ...@@ -20,6 +21,10 @@ task BgzipAndIndex {
File compressed = outputGz File compressed = outputGz
File index = outputGz + ".tbi" File index = outputGz + ".tbi"
} }
runtime {
docker: "quay.io/biocontainers/tabix:" + dockerTag
}
} }
task Index { task Index {
...@@ -185,6 +190,7 @@ task Tabix { ...@@ -185,6 +190,7 @@ task Tabix {
input { input {
String inputFile String inputFile
String type = "vcf" String type = "vcf"
String dockerTag = "0.2.6--ha92aebf_0"
} }
command { command {
...@@ -194,6 +200,10 @@ task Tabix { ...@@ -194,6 +200,10 @@ task Tabix {
output { output {
File index = inputFile + ".tbi" File index = inputFile + ".tbi"
} }
runtime {
docker: "quay.io/biocontainers/tabix:" + dockerTag
}
} }
task View { task View {
......
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment