Add mpileup subworkflow for paired, germline and tumoronly varcalling

.github/workflows/ci.yml

@@ -42,6 +42,7 @@ jobs:
           - "haplotypecaller"
           - "manta"
           - "markduplicates"
+          - "mpileup"
           - "msisensorpro"
           - "mutect2"
           - "prepare_recalibration"

CHANGELOG.md

@@ -33,6 +33,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
+- [#570](https://github.com/nf-core/sarek/pull/570) - Extract mpileup into its own subworkflow; zip mpileup files
 - [#383](https://github.com/nf-core/sarek/pull/383), [#528](https://github.com/nf-core/sarek/pull/528) - Update `CHANGELOG`
 - [#390](https://github.com/nf-core/sarek/pull/390) - Update `nextflow_schema.json`
 - [#408](https://github.com/nf-core/sarek/pull/408) - Sync `TEMPLATE` with `tools` `2.0.1`

conf/modules.config

@@ -682,11 +682,11 @@ process{
         ]
     }
 
-    withName: 'CAT_MPILEUP_.*' {
+    withName: 'CAT_MPILEUP' {
         publishDir       = [
             mode: params.publish_dir_mode,
-            path: { "${params.outdir}/variant_calling/${meta.id}/controlfreec" },
-            pattern: "*mpileup"
+            path: { "${params.outdir}/variant_calling/${meta.id}/mpileup" },
+            pattern: "*{mpileup.gz}",
         ]
     }
 
@@ -758,14 +758,13 @@ process{
         ]
     }
 
-    withName: 'MPILEUP_.*' {
-        ext.prefix       = { meta.num_intervals <= 1 ? meta.id : "${meta.id}_${intervals.simpleName}"}
-        ext.when         = { params.tools && params.tools.contains('controlfreec') }
+    withName: 'SAMTOOLS_MPILEUP' {
+        ext.when         = { params.tools && (params.tools.contains('controlfreec') || params.tools.contains('mpileup')) }
         publishDir       = [
             mode: params.publish_dir_mode,
-            path: { "${params.outdir}/variant_calling/${meta.id}/" },
-            pattern: "*mpileup",
-            saveAs: { meta.num_intervals > 1 ? null : "controlfreec/${it}" }
+            path: { "${params.outdir}/variant_calling/${meta.id}/mpileup" },
+            pattern: "*mpileup.gz",
+            saveAs: { meta.num_intervals > 1 ? null : it }
         ]
     }
 
@@ -910,20 +909,36 @@ process{
     }
 
     //CONTROLFREEC
-    withName: 'MPILEUP_NORMAL' {
-        ext.prefix       = { meta.num_intervals <= 1 ? "${meta.id}.normal" :  "${meta.id}_${intervals.simpleName}.normal" }
+    withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_MPILEUP_NORMAL:SAMTOOLS_MPILEUP' {
+        ext.prefix = { meta.num_intervals <= 1 ? "${meta.id}.normal" : "${meta.id}_${intervals.simpleName}.normal" }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_MPILEUP_TUMOR:SAMTOOLS_MPILEUP' {
+        ext.prefix = { meta.num_intervals <= 1 ? "${meta.id}.tumor" : "${meta.id}_${intervals.simpleName}.tumor" }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_MPILEUP_NORMAL:CAT_MPILEUP' {
+        ext.prefix = { "${meta.id}.normal.mpileup.gz" }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_MPILEUP_TUMOR:CAT_MPILEUP' {
+        ext.prefix = { "${meta.id}.tumor.mpileup.gz" }
     }
 
-    withName: 'MPILEUP_TUMOR' {
-        ext.prefix       = { meta.num_intervals <= 1 ? "${meta.id}.tumor" :  "${meta.id}_${intervals.simpleName}.tumor" }
+    withName: 'NFCORE_SAREK:SAREK:TUMOR_ONLY_VARIANT_CALLING:RUN_MPILEUP:CAT_MPILEUP' {
+        ext.prefix = { "${meta.id}.tumor.mpileup.gz" }
     }
 
-    withName: 'CAT_MPILEUP_NORMAL' {
-        ext.prefix = { "${meta.id}.normal.mpileup" }
+    withName: 'NFCORE_SAREK:SAREK:GERMLINE_VARIANT_CALLING:RUN_MPILEUP:SAMTOOLS_MPILEUP' {
+        ext.prefix = { meta.num_intervals <= 1 ? "${meta.id}.normal" : "${meta.id}_${intervals.simpleName}.normal" }
     }
 
-    withName: 'CAT_MPILEUP_TUMOR' {
-        ext.prefix = { "${meta.id}.tumor.mpileup" }
+    withName: 'NFCORE_SAREK:SAREK:GERMLINE_VARIANT_CALLING:RUN_MPILEUP:CAT_MPILEUP' {
+        ext.prefix = { "${meta.id}.normal.mpileup.gz" }
+    }
+
+    withName: 'NFCORE_SAREK:SAREK:TUMOR_ONLY_VARIANT_CALLING:RUN_MPILEUP:SAMTOOLS_MPILEUP' {
+        ext.prefix = { meta.num_intervals <= 1 ? "${meta.id}.tumor" : "${meta.id}_${intervals.simpleName}.tumor" }
     }
 
     withName: 'FREEC_SOMATIC'{
@@ -972,6 +987,7 @@ process{
             pattern: "*{bed,cnn,cnr,cns,pdf,png}"
         ]
     }
+
     //FREEBAYES
     withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_FREEBAYES_SOMATIC:FREEBAYES' {
         ext.args       = "--pooled-continuous \

modules.json

@@ -205,7 +205,7 @@
                 "git_sha": "897c33d5da084b61109500ee44c01da2d3e4e773"
             },
             "samtools/mpileup": {
-                "git_sha": "897c33d5da084b61109500ee44c01da2d3e4e773"
+                "git_sha": "24e05f6097a5dde57dd80d33295ed120f1b81aef"
             },
             "samtools/stats": {
                 "git_sha": "897c33d5da084b61109500ee44c01da2d3e4e773"

subworkflows/local/germline_variant_calling.nf

@@ -7,6 +7,7 @@ include { RUN_DEEPVARIANT     } from '../nf-core/variantcalling/deepvariant/main
 include { RUN_FREEBAYES       } from '../nf-core/variantcalling/freebayes/main.nf'
 include { RUN_HAPLOTYPECALLER } from '../nf-core/variantcalling/haplotypecaller/main.nf'
 include { RUN_MANTA_GERMLINE  } from '../nf-core/variantcalling/manta/germline/main.nf'
+include { RUN_MPILEUP         } from '../nf-core/variantcalling/mpileup/main'
 include { RUN_STRELKA_SINGLE  } from '../nf-core/variantcalling/strelka/single/main.nf'
 include { RUN_TIDDIT          } from '../nf-core/variantcalling/tiddit/main.nf'
 
@@ -39,6 +40,7 @@ workflow GERMLINE_VARIANT_CALLING {
     manta_vcf           = Channel.empty()
     strelka_vcf         = Channel.empty()
     tiddit_vcf          = Channel.empty()
+    mpileup             = Channel.empty()
 
     // Remap channel with intervals
     cram_recalibrated_intervals = cram_recalibrated.combine(intervals)
@@ -63,6 +65,18 @@ workflow GERMLINE_VARIANT_CALLING {
             cram, crai, bed_new, tbi_new]
         }
 
+    if(params.tools.contains('mpileup')){
+        cram_intervals_no_index = cram_recalibrated_intervals
+            .map { meta, cram, crai, intervals ->
+                [meta, cram, intervals]
+            }
+
+        RUN_MPILEUP(cram_intervals_no_index,
+                        fasta)
+        mpileup_germline = RUN_MPILEUP.out.mpileup
+        ch_versions = ch_versions.mix(RUN_MPILEUP.out.versions)
+    }
+
     // CNVKIT
 
     if(tools.contains('cnvkit')){

subworkflows/local/pair_variant_calling.nf

@@ -8,6 +8,8 @@ include { RUN_FREEBAYES as RUN_FREEBAYES_SOMATIC    } from '../nf-core/variantca
 include { RUN_MANTA_SOMATIC                         } from '../nf-core/variantcalling/manta/somatic/main.nf'
 include { RUN_STRELKA_SOMATIC                       } from '../nf-core/variantcalling/strelka/somatic/main.nf'
 include { RUN_CNVKIT_SOMATIC                        } from '../nf-core/variantcalling/cnvkit/somatic/main.nf'
+include { RUN_MPILEUP as RUN_MPILEUP_NORMAL         } from '../nf-core/variantcalling/mpileup/main'
+include { RUN_MPILEUP as RUN_MPILEUP_TUMOR          } from '../nf-core/variantcalling/mpileup/main'
 
 workflow PAIR_VARIANT_CALLING {
     take:
@@ -73,9 +75,17 @@ workflow PAIR_VARIANT_CALLING {
                     .map {meta, normal_cram, normal_crai, tumor_cram, tumor_crai, intervals ->
                             [meta, tumor_cram, intervals]
                         }
+        RUN_MPILEUP_NORMAL(cram_normal_intervals_no_index, fasta)
+        mpileup_normal = RUN_MPILEUP_NORMAL.out.mpileup
+        RUN_MPILEUP_TUMOR(cram_tumor_intervals_no_index, fasta)
+        mpileup_tumor = RUN_MPILEUP_TUMOR.out.mpileup
+
+        controlfreec_input = mpileup_normal.cross(mpileup_tumor)
+        .map{ normal, tumor ->
+            [normal[0], normal[1], tumor[1], [], [], [], []]
+        }
 
-        RUN_CONTROLFREEC_SOMATIC(cram_normal_intervals_no_index,
-                        cram_tumor_intervals_no_index,
+        RUN_CONTROLFREEC_SOMATIC(controlfreec_input,
                         fasta,
                         fasta_fai,
                         dbsnp,

subworkflows/local/tumor_variant_calling.nf

@@ -9,6 +9,7 @@ include { RUN_MANTA_TUMORONLY                     } from '../nf-core/variantcall
 include { RUN_STRELKA_SINGLE                      } from '../nf-core/variantcalling/strelka/single/main.nf'
 include { RUN_CONTROLFREEC_TUMORONLY              } from '../nf-core/variantcalling/controlfreec/tumoronly/main.nf'
 include { RUN_CNVKIT_TUMORONLY                    } from '../nf-core/variantcalling/cnvkit/tumoronly/main.nf'
+include { RUN_MPILEUP                             } from '../nf-core/variantcalling/mpileup/main'
 
 workflow TUMOR_ONLY_VARIANT_CALLING {
     take:
@@ -62,12 +63,22 @@ workflow TUMOR_ONLY_VARIANT_CALLING {
             cram, crai, bed_new, tbi_new]
         }
 
-    if(tools.contains('controlfreec')){
+    if (tools.contains('mpileup') || tools.contains('controlfreec')){
         cram_intervals_no_index = cram_recalibrated_intervals.map { meta, cram, crai, intervals ->
                                                                     [meta, cram, intervals]
                                                                     }
+        RUN_MPILEUP(cram_intervals_no_index,
+                        fasta)
+    }
+
+    if (tools.contains('controlfreec')){
+        controlfreec_input = RUN_MPILEUP.out.mpileup
+                                .map{ meta, pileup_tumor ->
+                                    [meta, [], pileup_tumor, [], [], [], []]
+                                }
+
         RUN_CONTROLFREEC_TUMORONLY(
-                        cram_intervals_no_index,
+                        controlfreec_input,
                         fasta,
                         fasta_fai,
                         dbsnp,