Dev manta

ControlFREEC/11.5/AssessSignificance.nf

@@ -0,0 +1,18 @@
+process AssessSignificance {
+    tag {"Control Freec AssessSignificance ${sample_id}"}
+    label 'ControlFreec_11_5'
+    label 'ControlFreec_11_5_AssessSignificance'
+    container = 'library://sawibo/default/bioinf-tools:freec11.5'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(ratio_file), path(cnv_file))
+
+    output:
+        tuple(sample_id, path("${cnv_file.name}.p.value.txt"), emit: cnv_pvalue)
+
+    script:
+        """
+        cat /bin/assess_significance.R | R --slave --args ${cnv_file} ${ratio_file}
+        """
+}

ControlFREEC/11.5/Freec.nf

@@ -0,0 +1,36 @@
+process Freec {
+    tag {"Control Freec ${sample_id}"}
+    label 'ControlFreec_11_5'
+    label 'ControlFreec_11_5_Freec'
+    //TODO: upload to singularity library
+    container = 'library://sawibo/default/bioinf-tools:freec11.5'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(bam_file), path(bai_file))
+
+    output:
+        tuple(sample_id, path("${bam_file.name}_ratio.txt"), path("${bam_file.name}_CNVs"), emit: cnv)
+        tuple(sample_id, path("${bam_file.name}_sample.cpn"), path("${bam_file.name}_ratio.BedGraph"), path("${bam_file.name}_info.txt"), emit: other)
+
+    script:
+        def config = "${sample_id}.config"
+        """
+        touch ${config}
+        echo "[general]" >> ${config}
+        echo "chrLenFile = ${params.chr_len_file}" >> ${config}
+        echo "chrFiles = ${params.chr_files}" >> ${config}
+        echo "gemMappabilityFile = ${params.gem_mappability_file}" >> ${config}
+        echo "ploidy = ${params.ploidy}" >> ${config}
+        echo "window = ${params.window}" >> ${config}
+        echo "telocentromeric = ${params.telocentromeric}" >> ${config}
+        echo "BedGraphOutput=TRUE" >> ${config}
+        echo "maxThreads=${task.cpus}" >> ${config}
+
+        echo "[sample]" >> ${config}
+        echo "inputFormat = BAM" >> ${config}
+        echo "mateFile = ${bam_file}" >> ${config}
+
+        freec -conf ${config}
+        """
+}

ControlFREEC/11.5/MakeGraph.nf

@@ -0,0 +1,18 @@
+process MakeGraph {
+    tag {"Control Freec MakeGraph ${sample_id}"}
+    label 'ControlFreec_11_5'
+    label 'ControlFreec_11_5_MakeGraph'
+    container = 'library://sawibo/default/bioinf-tools:freec11.5'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(ratio_file), path(cnv_file))
+
+    output:
+        tuple(sample_id, path("${ratio_file.name}.png"), path("${ratio_file.name}.log2.png"), emit: ratio_png)
+
+    script:
+        """
+        cat /bin/makeGraph.R | R --slave --args ${params.ploidy} ${ratio_file}
+        """
+}

ControlFREEC/11.5/MakeKaryotype.nf

@@ -0,0 +1,18 @@
+process MakeKaryotype {
+    tag {"Control Freec MakeKaryotype ${sample_id}"}
+    label 'ControlFreec_11_5'
+    label 'ControlFreec_11_5_MakeKaryotype'
+    container = 'library://sawibo/default/bioinf-tools:freec11.5'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(ratio_file), path(cnv_file))
+
+    output:
+        tuple(sample_id, path("*_karyotype.pdf"), emit: karyotype_pdf)
+
+    script:
+        """
+        cat /bin/makeKaryotype.R | R --slave --args ${params.ploidy} ${params.maxlevel} ${params.telocentromeric} ${ratio_file}
+        """
+}

GATK/3.8-1-0-gf15c1c3ef/BaseRecalibrator.nf

@@ -0,0 +1,32 @@
+process BaseRecalibrator {
+    tag {"GATK BaseRecalibrator ${sample_id} - ${chr}"}
+    label 'GATK_3_8_1_0_gf15c1c3ef'
+    label 'GATK_3_8_1_0_gf15c1c3ef_BaseRecalibrator'
+    container = 'quay.io/biocontainers/gatk:3.8--py27_1'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(bam_file), path(bai_file), chr)
+
+    output:
+        tuple(sample_id, path("${bam_file.baseName}.bqsr.${chr}.bam"), path("${bam_file.baseName}.bqsr.${chr}.bai"), emit: bam_file)
+
+    script:
+        """
+        java -Xmx${task.memory.toGiga()-4}G -jar $params.gatk_path -T BaseRecalibrator \
+        --num_cpu_threads_per_data_thread ${task.cpus} \
+        --reference_sequence ${params.genome} \
+        --input_file ${bam_file} \
+        --intervals ${chr} \
+        --out ${bam_file.baseName}.bqsr.${chr}.table \
+        ${params.optional}
+
+        java -Xmx${task.memory.toGiga()-4}G -jar ${params.gatk_path} -T PrintReads \
+        --num_cpu_threads_per_data_thread ${task.cpus} \
+        --reference_sequence ${params.genome} \
+        --input_file ${bam_file} \
+        --BQSR ${bam_file.baseName}.bqsr.${chr}.table \
+        --intervals ${chr} \
+        --out ${bam_file.baseName}.bqsr.${chr}.bam
+        """
+}

GATK/3.8-1-0-gf15c1c3ef/CatVariants.nf

@@ -0,0 +1,19 @@
+process CatVariantsGVCF {
+    tag {"GATK CatVariantsGVCF ${sample_id}"}
+    label 'GATK_3_8_1_0_gf15c1c3ef'
+    label 'GATK_3_8_1_0_gf15c1c3ef_CatVariantsGVCF'
+    container = 'quay.io/biocontainers/gatk:3.8--py27_1'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(gvcf_files), path(gvcf_idx_files))
+
+    output:
+        tuple(sample_id, path("${sample_id}.g.vcf.gz"), path("${sample_id}.g.vcf.gz.tbi"), emit:vcf_file)
+
+    script:
+        def input_files = gvcf_files.collect{"$it"}.join(" -V ")
+        """
+        java -Xmx${task.memory.toGiga()-4}G -cp ${params.gatk_path} org.broadinstitute.gatk.tools.CatVariants --reference ${params.genome} -V ${input_files} --outputFile ${sample_id}.g.vcf.gz ${params.optional}
+        """
+}

GATK/3.8-1-0-gf15c1c3ef/CombineVariants.nf

@@ -17,3 +17,23 @@ process CombineVariants {
         java -Xmx${task.memory.toGiga()-4}G -jar ${params.gatk_path} -T CombineVariants --reference_sequence ${params.genome} -V ${input_files} --out ${analysis_id}.vcf ${params.optional}
         """
 }
+
+process CombineVariantsGVCF {
+    tag {"GATK CombineVariantsGVCF ${sample_id}"}
+    label 'GATK_3_8_1_0_gf15c1c3ef'
+    label 'GATK_3_8_1_0_gf15c1c3ef_CombineVariantsGVCF'
+    container = 'quay.io/biocontainers/gatk:3.8--py27_1'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(vcf_files), path(vcf_idx_files))
+
+    output:
+        tuple(sample_id, path("${sample_id}.g.vcf"), path("${sample_id}.g.vcf.idx"), emit:vcf_file)
+
+    script:
+        def input_files = vcf_files.collect{"$it"}.join(" -V ")
+        """
+        java -Xmx${task.memory.toGiga()-4}G -jar ${params.gatk_path} -T CombineVariants --reference_sequence ${params.genome} -V ${input_files} --out ${sample_id}.g.vcf ${params.optional}
+        """
+}

GATK/3.8-1-0-gf15c1c3ef/GenotypeGVCFs.nf

@@ -0,0 +1,24 @@
+process GenotypeGVCFs {
+    tag {"GATK GenotypeGVCFs ${analysis_id}"}
+    label 'GATK_3_8_1_0_gf15c1c3ef'
+    label 'GATK_3_8_1_0_gf15c1c3ef_GenotypeGVCFs'
+    container = 'quay.io/biocontainers/gatk:3.8--py27_1'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(analysis_id, path(gvcf_files), path(gvcf_idx_files), path(interval_file))
+
+    output:
+        tuple(analysis_id, path("${analysis_id}_${interval_file.baseName}.vcf"), path("${analysis_id}_${interval_file.baseName}.vcf.idx"), emit:vcf_file)
+
+    script:
+        def input_files = gvcf_files.collect{"$it"}.join(" -V ")
+        """
+        java -Xmx${task.memory.toGiga()-4}G -jar ${params.gatk_path} -T GenotypeGVCFs \
+        --reference_sequence ${params.genome} \
+        -V ${input_files} \
+        --out ${analysis_id}_${interval_file.baseName}.vcf \
+        --intervals ${interval_file} \
+        ${params.optional} \
+        """
+}

GATK/3.8-1-0-gf15c1c3ef/HaplotypeCaller.nf

@@ -9,7 +9,7 @@ process HaplotypeCaller {
         tuple(analysis_id, path(bam_files), path(bai_files), path(interval_file))
 
     output:
-        tuple(val(analysis_id), file("${analysis_id}.${interval_file.baseName}.vcf"), file("${analysis_id}.${interval_file.baseName}.vcf.idx"), emit: vcf_file)
+        tuple(val(analysis_id), path("${analysis_id}.${interval_file.baseName}.vcf"), path("${analysis_id}.${interval_file.baseName}.vcf.idx"), emit: vcf_file)
 
     script:
         def input_files = bam_files.collect{"$it"}.join(" --input_file ")
@@ -22,3 +22,28 @@ process HaplotypeCaller {
         ${params.optional}
         """
 }
+
+process HaplotypeCallerGVCF {
+    tag {"GATK HaplotypeCallerGVCF ${sample_id} - ${interval_file.baseName}"}
+    label 'GATK_3_8_1_0_gf15c1c3ef'
+    label 'GATK_3_8_1_0_gf15c1c3ef_HaplotypeCallerGVCF'
+    container = 'quay.io/biocontainers/gatk:3.8--py27_1'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(bam_file), path(bai_file), path(interval_file))
+
+    output:
+        tuple(val(sample_id), path("${sample_id}_${interval_file.baseName}.g.vcf"), path("${sample_id}_${interval_file.baseName}.g.vcf.idx"), path(interval_file), emit: vcf_file)
+
+    script:
+        """
+        java -Xmx${task.memory.toGiga()-4}G -jar ${params.gatk_path} -T HaplotypeCaller \
+        --reference_sequence ${params.genome} \
+        --input_file ${bam_file} \
+        --intervals ${interval_file} \
+        --out ${sample_id}_${interval_file.baseName}.g.vcf \
+        --emitRefConfidence GVCF \
+        ${params.optional}
+        """
+}

Manta/1.6.0/Manta.nf

@@ -0,0 +1,26 @@
+process Manta {
+    tag {"Manta ConfigAndRun ${sample_id}"}
+    label 'Manta_1_6_0'
+    label 'Manta_1_6_0_ConfigAndRun'
+    container = 'quay.io/biocontainers/manta:1.6.0--py27_0'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple (sample_id, path(bam_file), path(bai_file))
+
+    output:
+        tuple (sample_id, path("*.candidateSmallIndels.*"),path("*.candidateSV.*"),path("*.diploidSV.*"), emit: sv )
+
+    script:
+        """
+        configManta.py --referenceFasta ${params.genome_fasta} --runDir . --bam $bam_file
+        ./runWorkflow.py -m local -j ${task.cpus}
+
+        mv results/variants/candidateSmallIndels.vcf.gz Manta_${sample_id}.candidateSmallIndels.vcf.gz
+        mv results/variants/candidateSmallIndels.vcf.gz.tbi Manta_${sample_id}.candidateSmallIndels.vcf.gz.tbi
+        mv results/variants/candidateSV.vcf.gz Manta_${sample_id}.candidateSV.vcf.gz
+        mv results/variants/candidateSV.vcf.gz.tbi Manta_${sample_id}.candidateSV.vcf.gz.tbi
+        mv results/variants/diploidSV.vcf.gz Manta_${sample_id}.diploidSV.vcf.gz
+        mv results/variants/diploidSV.vcf.gz.tbi Manta_${sample_id}.diploidSV.vcf.gz.tbi
+        """
+}

MultiQC/1.9/MultiQC.nf

@@ -0,0 +1,18 @@
+process MultiQC {
+    tag {"MultiQC"}
+    label 'MultiQC_1_9'
+    container = 'quay.io/biocontainers/multiqc:1.9--pyh9f0ad1d_0'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        val(analysis_id)
+        path(qc_files)
+
+    output:
+        tuple(path("${analysis_id}_multiqc_report.html"), path("${analysis_id}_multiqc_report_data"), emit: report)
+
+    script:
+        """
+        multiqc ${params.optional} --title ${analysis_id} .
+        """
+}

Picard/2.22.0/CollectWgsMetrics.nf

@@ -0,0 +1,18 @@
+process CollectWgsMetrics {
+    tag {"PICARD CollectWgsMetrics ${sample_id}"}
+    label 'PICARD_2_22_0'
+    label 'PICARD_2_22_0_CollectWgsMetrics'
+    container = 'quay.io/biocontainers/picard:2.22.0--0'
+    shell = ['/bin/bash', '-euo', 'pipefail']
+
+    input:
+        tuple(sample_id, path(bam_file), path(bai_file))
+
+    output:
+        path("${sample_id}.wgs_metrics.txt", emit: txt_file)
+
+    script:
+        """
+        picard -Xmx${task.memory.toGiga()-4}G CollectWgsMetrics TMP_DIR=\$TMPDIR R=${params.genome} INPUT=${bam_file} OUTPUT=${sample_id}.wgs_metrics.txt ${params.optional}
+        """
+}

Picard/2.22.0/IntervalListTools.nf

@@ -18,7 +18,8 @@ process IntervalListTools {
         SUBDIVISION_MODE=BALANCING_WITHOUT_INTERVAL_SUBDIVISION_WITH_OVERFLOW \
         SCATTER_COUNT=${params.scatter_count} \
         UNIQUE=true \
-
+        ${params.optional}
+
         for folder in temp*; do mv \$folder/scattered.interval_list \$folder/\$folder\\.interval_list; done
         """
 }