Sambamba

modules.json

@@ -339,6 +339,11 @@
  "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",
  "installed_by": ["bam_ngscheckmate"]
  },
+ "sambamba/markdup": {
+ "branch": "master",
+ "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",
+ "installed_by": ["modules"]
+ },
  "samblaster": {
  "branch": "master",
  "git_sha": "3f5420aa22e00bd030a2556dfdffc9e164ec0ec5",

subworkflows/local/bam_markduplicates_sambamba/main.nf

@@ -0,0 +1,48 @@
+//
+// MARKDUPLICATES AND QC after mapping
+//
+// For all modules here:
+// A when clause condition is defined in the conf/modules.config to determine if the module should be run
+
+include { CRAM_QC_MOSDEPTH_SAMTOOLS } from '../cram_qc_mosdepth_samtools/main'
+include { SAMBAMBA_MARKDUP } from '../../../modules/nf-core/sambamba/markdup/main'
+
+workflow BAM_MARKDUPLICATES_SAMBAMBA {
+ take:
+ bam // channel: [mandatory] [ meta, bam ]
+ fasta // channel: [mandatory] [ fasta ]
+ fasta_fai // channel: [mandatory] [ fasta_fai ]
+ intervals_bed_combined // channel: [optional] [ intervals_bed ]
+
+ main:
+ versions = Channel.empty()
+ reports = Channel.empty()
+
+ // RUN MARKUPDUPLICATES
+ bam.view()
+ SAMBAMBA_MARKDUP(bam)
+
+ // bam = SAMBAMBA_MARKDUP.out.bam
+ // bam.view()
+
+ cram = bam
+ // Join with the crai file
+ // bam = SAMBAMBA_MARKDUP.out.bam.join(SAMBAMBA_MARKDUPLICATES.out.bai, failOnDuplicate: true, failOnMismatch: true)
+
+ // QC on CRAM
+ // CRAM_QC_MOSDEPTH_SAMTOOLS(bam, fasta, intervals_bed_combined)
+
+ // Gather all reports generated
+ // reports = reports.mix(SAMBAMBA_MARKDUP.out.metrics)
+ // reports = reports.mix(CRAM_QC_MOSDEPTH_SAMTOOLS.out.reports)
+
+ // Gather versions of all tools used
+ versions = versions.mix(SAMBAMBA_MARKDUP.out.versions)
+ // versions = versions.mix(CRAM_QC_MOSDEPTH_SAMTOOLS.out.versions)
+
+ emit:
+ cram
+ reports
+
+ versions // channel: [ versions.yml ]
+}

workflows/sarek.nf

@@ -185,6 +185,7 @@ include { SAMTOOLS_CONVERT as CRAM_TO_BAM_RECAL } from '../modules/nf-core
 
 // Mark Duplicates (+QC)
 include { BAM_MARKDUPLICATES } from '../subworkflows/local/bam_markduplicates/main'
+include { BAM_MARKDUPLICATES_SAMBAMBA } from '../subworkflows/local/bam_markduplicates_sambamba/main'
 include { BAM_MARKDUPLICATES_SPARK } from '../subworkflows/local/bam_markduplicates_spark/main'
 include { BAM_SENTIEON_DEDUP } from '../subworkflows/local/bam_sentieon_dedup/main'
 
@@ -611,7 +612,24 @@ workflow SAREK {
 
  // Gather used softwares versions
  versions = versions.mix(BAM_SENTIEON_DEDUP.out.versions)
- } else {
+ } else if (params.sambamba_dedup) {
+ BAM_MARKDUPLICATES_SAMBAMBA(
+ cram_for_markduplicates,
+ fasta,
+ fasta_fai,
+ intervals_for_preprocessing)
+
+ cram_markduplicates_no_spark = BAM_MARKDUPLICATES_SAMBAMBA.out.cram
+
+ // Gather QC reports
+ reports = reports.mix(BAM_MARKDUPLICATES_SAMBAMBA.out.reports.collect{ meta, report -> report })
+
+ // Gather used softwares versions
+ versions = versions.mix(BAM_MARKDUPLICATES_SAMBAMBA.out.versions)
+ }
+
+
+ else {
  BAM_MARKDUPLICATES(
  cram_for_markduplicates,
  fasta,