feat: integrate Bracken into the profiling pipeline

conf/modules.config

@@ -277,7 +277,7 @@ process {
     }
 
     withName: KRAKEN2_KRAKEN2 {
-        ext.args = { "${meta.db_params}" }
+        ext.args = params.kraken2_save_minimizers ? { "${meta.db_params} --report-minimizer-data" } : { "${meta.db_params}" }
         ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
         publishDir = [
             path: { "${params.outdir}/kraken2/${meta.db_name}/" },
@@ -286,6 +286,15 @@ process {
         ]
     }
 
+    withName: BRACKEN_BRACKEN {
+        errorStrategy = 'ignore'
+        publishDir = [
+            path: { "${params.outdir}/bracken/${meta.db_name}/" },
+            mode: params.publish_dir_mode,
+            pattern: '*.tsv'
+        ]
+    }
+
     withName: KRAKENTOOLS_COMBINEKREPORTS {
         ext.prefix = { "kraken2_${meta.id}_combined_reports" }
         publishDir = [

modules/local/kraken_standard_report.nf

@@ -1,13 +1,11 @@
 process KRAKEN_STANDARD_REPORT {
     tag "$meta.id"
-    label 'process_low'
+    label 'process_single'
 
     conda (params.enable_conda ? 'conda-forge::sed=4.8' : null)
-    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        container 'https://containers.biocontainers.pro/s3/SingImgsRepo/biocontainers/v1.2.0_cv2/biocontainers_v1.2.0_cv2.img'
+        'https://containers.biocontainers.pro/s3/SingImgsRepo/biocontainers/v1.2.0_cv2/biocontainers_v1.2.0_cv2.img' :
-    } else {
+        'biocontainers/biocontainers:v1.2.0_cv2' }"
-        container 'biocontainers/biocontainers:v1.2.0_cv2'
-    }
 
     input:
     tuple val(meta), path(report)
@@ -15,10 +13,14 @@ process KRAKEN_STANDARD_REPORT {
     output:
     tuple val(meta), path(result), emit: report
 
+    when:
+    task.ext.when == null || task.ext.when
+
     script:
-    result = "${report.baseName}_standardized.kraken2.report.txt"
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    result = "${prefix}_standardized.kraken2.report.txt"
     """
-    cut -f1-3,6-8 "${report}" > "${result}"
+    cut -f1-3,6-8 '${report}' > '${result}'
     """
 }
 

nextflow.config

@@ -114,6 +114,10 @@ params {
     run_kraken2                     = false
     kraken2_save_reads              = false // added directly to module in profiling.nf
     kraken2_save_readclassification = false // added directly to module in profiling.nf
+    kraken2_save_minimizers         = false
+
+    // Bracken
+    run_bracken = true
 
     // centrifuge
     run_centrifuge             = false

subworkflows/local/profiling.nf

@@ -5,6 +5,7 @@
 include { MALT_RUN                              } from '../../modules/nf-core/malt/run/main'
 include { MEGAN_RMA2INFO as MEGAN_RMA2INFO_TSV  } from '../../modules/nf-core/megan/rma2info/main'
 include { KRAKEN2_KRAKEN2                       } from '../../modules/nf-core/kraken2/kraken2/main'
+include { BRACKEN_BRACKEN                       } from '../../modules/nf-core/bracken/bracken/main'
 include { CENTRIFUGE_CENTRIFUGE                 } from '../../modules/nf-core/centrifuge/centrifuge/main'
 include { CENTRIFUGE_KREPORT                    } from '../../modules/nf-core/centrifuge/kreport/main'
 include { METAPHLAN3_METAPHLAN3                 } from '../../modules/nf-core/metaphlan3/metaphlan3/main'
@@ -133,6 +134,33 @@ workflow PROFILING {
 
     }
 
+    if ( params.run_kraken2 && params.run_bracken ) {
+
+        def ch_input_for_bracken
+
+        if (params.kraken2_save_minimizers) {
+            ch_input_for_bracken = KRAKEN_STANDARD_REPORT(KRAKEN2_KRAKEN2.out.report).report
+        } else {
+            ch_input_for_bracken = KRAKEN2_KRAKEN2.out.report
+        }
+
+        ch_input_for_bracken = ch_input_for_bracken
+            .combine(
+                databases.filter { meta, db ->
+                    meta['tool'] == 'bracken'
+                }
+            )
+            .multiMap { meta, report, db_meta, db ->
+                report: [meta + db_meta, report]
+                db: db
+            }
+
+        BRACKEN_BRACKEN(ch_input_for_bracken.report, ch_input_for_bracken.db)
+        ch_versions     = ch_versions.mix(BRACKEN_BRACKEN.out.versions.first())
+        ch_raw_profiles = ch_raw_profiles.mix(BRACKEN_BRACKEN.out.reports)
+
+    }
+
     if ( params.run_centrifuge ) {
 
         ch_input_for_centrifuge =  ch_input_for_profiling.centrifuge