Merge pull request #642 from FriederikeHanssen/ref_files

Small ref file issues for ASCAT, Controlfreec, & CNVKIT
nf-core · Jul 18, 2022 · 2614122 · 2614122
2 parents c43ef33 + ac22900
commit 2614122
Show file tree

Hide file tree

Showing 10 changed files with 90 additions and 155 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -138,6 +138,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [#618](https://github.com/nf-core/sarek/pull/618) - Fix channel issue with `targets.bed` in prepare_intervals
 - [#634](https://github.com/nf-core/sarek/pull/634) - Fix issue with samtools/mosdepth plots in multiqc_report
 - [#641](https://github.com/nf-core/sarek/pull/641) - Fix issue with duplicate substring in tools and skip_tools
+- [#642](https://github.com/nf-core/sarek/pull/642) - Only unzip ref files if tool is run, only publish ref files if `--save_reference` and simplify CNKit logic
 - [#650](https://github.com/nf-core/sarek/pull/650) - Fix intervals checks
 
 ### Deprecated

diff --git a/conf/modules.config b/conf/modules.config
@@ -130,6 +130,36 @@ process {
         ]
     }
 
+    withName: 'UNZIP_ALLELES|UNZIP_LOCI|UNZIP_GC|UNZIP_RT'{
+        ext.when         = { params.tools && params.tools.split(',').contains('ascat')}
+    }
+
+    withName: 'UNTAR_CHR_DIR'{
+        ext.when         = { params.tools && params.tools.split(',').contains('controlfreec')}
+    }
+
+    withName: 'CNVKIT_ANTITARGET' {
+        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
+        publishDir       = [
+            enabled: params.save_reference,
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/reference" },
+            pattern: "*{bed}"
+        ]
+    }
+
+    withName: 'CNVKIT_REFERENCE' {
+        ext.prefix       = "cnvkit"
+        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
+        publishDir       = [
+            enabled: params.save_reference,
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/reference" },
+            pattern: "*{cnn}"
+        ]
+    }
+
+
 // PREPARE INTERVALS
 
     withName: 'CREATE_INTERVALS_BED' {
@@ -519,8 +549,8 @@ process {
 // VARIANT CALLING
 process{
 
-    // CNVKIT_GERMLINE
-    withName: 'CNVKIT_BATCH_GERMLINE' {
+    // CNVKIT
+    withName: 'CNVKIT_BATCH' {
         ext.args         = { params.wes ? "--method hybrid --diagram --scatter" : "--method wgs --diagram --scatter" }
         ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
         publishDir       = [
@@ -817,36 +847,6 @@ process{
         ]
     }
 
-    //CNVKIT
-    withName: 'CNVKIT_ANTITARGET' {
-        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
-        publishDir       = [
-            mode: params.publish_dir_mode,
-            path: { "${params.outdir}/reference" },
-            pattern: "*{bed}"
-        ]
-    }
-
-    withName: 'CNVKIT_REFERENCE' {
-        ext.prefix       = "cnvkit"
-        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
-        publishDir       = [
-            mode: params.publish_dir_mode,
-            path: { "${params.outdir}/reference" },
-            pattern: "*{cnn}"
-        ]
-    }
-
-    withName: 'CNVKIT_BATCH_TUMORONLY' {
-        ext.args         = { params.wes ? "--method hybrid --diagram --scatter" : "--method wgs --diagram --scatter" }
-        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
-        publishDir       = [
-            mode: params.publish_dir_mode,
-            path: { "${params.outdir}/variant_calling/${meta.id}/cnvkit" },
-            pattern: "*{bed,cnn,cnr,cns,pdf,png}"
-        ]
-    }
-
     //MANTA
     withName: 'MERGE_MANTA_TUMOR' {
         ext.prefix       = {"${meta.id}.manta.tumor_sv"}
@@ -1044,17 +1044,6 @@ process{
             }
     }
 
-    //CNVKIT
-    withName: 'CNVKIT_BATCH_SOMATIC' {
-        ext.args         = { params.wes ? "--method hybrid --diagram --scatter" : "--method wgs --diagram --scatter" }
-        ext.when         = { params.tools && params.tools.split(',').contains('cnvkit') }
-        publishDir       = [
-            mode: params.publish_dir_mode,
-            path: { "${params.outdir}/variant_calling/${meta.id}/cnvkit" },
-            pattern: "*{bed,cnn,cnr,cns,pdf,png}"
-        ]
-    }
-
     //FREEBAYES
     if (params.tools && params.tools.split(',').contains('freebayes')) {
         withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_FREEBAYES_SOMATIC:FREEBAYES' {

diff --git a/subworkflows/local/germline_variant_calling.nf b/subworkflows/local/germline_variant_calling.nf
@@ -2,7 +2,7 @@
 // GERMLINE VARIANT CALLING
 //
 
-include { RUN_CNVKIT_GERMLINE } from '../nf-core/variantcalling/cnvkit/germline/main.nf'
+include { RUN_CNVKIT          } from '../nf-core/variantcalling/cnvkit/main.nf'
 include { RUN_DEEPVARIANT     } from '../nf-core/variantcalling/deepvariant/main.nf'
 include { RUN_FREEBAYES       } from '../nf-core/variantcalling/freebayes/main.nf'
 include { RUN_HAPLOTYPECALLER } from '../nf-core/variantcalling/haplotypecaller/main.nf'
@@ -85,12 +85,12 @@ workflow GERMLINE_VARIANT_CALLING {
                 [meta, [], cram]
             }
 
-        RUN_CNVKIT_GERMLINE(cram_recalibrated_cnvkit_germline,
+        RUN_CNVKIT(cram_recalibrated_cnvkit_germline,
                             fasta,
                             fasta_fai,
                             intervals_bed_combined,
                             [])
-        ch_versions     = ch_versions.mix(RUN_CNVKIT_GERMLINE.out.versions)
+        ch_versions     = ch_versions.mix(RUN_CNVKIT.out.versions)
     }
 
     // DEEPVARIANT

diff --git a/subworkflows/local/pair_variant_calling.nf b/subworkflows/local/pair_variant_calling.nf
@@ -7,7 +7,7 @@ include { RUN_CONTROLFREEC_SOMATIC                  } from '../nf-core/variantca
 include { RUN_FREEBAYES as RUN_FREEBAYES_SOMATIC    } from '../nf-core/variantcalling/freebayes/main.nf'
 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_CNVKIT                                } from '../nf-core/variantcalling/cnvkit/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'
 include { RUN_ASCAT_SOMATIC                         } from '../nf-core/variantcalling/ascat/main'
@@ -130,11 +130,11 @@ workflow PAIR_VARIANT_CALLING {
                 [meta, tumor_cram, normal_cram]
             }
 
-        RUN_CNVKIT_SOMATIC( cram_pair_cnvkit_somatic,
-                            fasta,
-                            fasta_fai,
-                            intervals_bed_combined,
-                            [])
+        RUN_CNVKIT( cram_pair_cnvkit_somatic,
+                    fasta,
+                    fasta_fai,
+                    intervals_bed_combined,
+                    [])
     }
 
     if (tools.split(',').contains('freebayes')){

diff --git a/subworkflows/local/prepare_genome.nf b/subworkflows/local/prepare_genome.nf
@@ -10,6 +10,8 @@
 
 include { BWA_INDEX as BWAMEM1_INDEX             } from '../../modules/nf-core/modules/bwa/index/main'
 include { BWAMEM2_INDEX                          } from '../../modules/nf-core/modules/bwamem2/index/main'
+include {CNVKIT_ANTITARGET                       } from '../../modules/nf-core/modules/cnvkit/antitarget/main'
+include {CNVKIT_REFERENCE                        } from '../../modules/nf-core/modules/cnvkit/reference/main'
 include { DRAGMAP_HASHTABLE                      } from '../../modules/nf-core/modules/dragmap/hashtable/main'
 include { GATK4_CREATESEQUENCEDICTIONARY         } from '../../modules/nf-core/modules/gatk4/createsequencedictionary/main'
 include { MSISENSORPRO_SCAN                      } from '../../modules/nf-core/modules/msisensorpro/scan/main'
@@ -26,17 +28,18 @@ include { UNZIP as UNZIP_RT                      } from '../../modules/nf-core/m
 
 workflow PREPARE_GENOME {
     take:
-        ascat_alleles     // channel: [optional]  ascat allele files
-        ascat_loci        // channel: [optional]  ascat loci files
-        ascat_loci_gc     // channel: [optional]  ascat gc content file
-        ascat_loci_rt     // channel: [optional]  ascat replictiming file
-        chr_dir           // channel: [optional]  chromosome files
-        dbsnp             // channel: [optional]  dbsnp
-        fasta             // channel: [mandatory] fasta
-        fasta_fai         // channel: [optional]  fasta_fai
-        germline_resource // channel: [optional]  germline_resource
-        known_indels      // channel: [optional]  known_indels
-        pon               // channel: [optional]  pon
+        ascat_alleles           // channel: [optional]  ascat allele files
+        ascat_loci              // channel: [optional]  ascat loci files
+        ascat_loci_gc           // channel: [optional]  ascat gc content file
+        ascat_loci_rt           // channel: [optional]  ascat replictiming file
+        chr_dir                 // channel: [optional]  chromosome files
+        dbsnp                   // channel: [optional]  dbsnp
+        fasta                   // channel: [mandatory] fasta
+        fasta_fai               // channel: [optional]  fasta_fai
+        germline_resource       // channel: [optional]  germline_resource
+        intervals_bed_combined  // channel: []
+        known_indels            // channel: [optional]  known_indels
+        pon                     // channel: [optional]  pon
 
 
     main:
@@ -60,6 +63,10 @@ workflow PREPARE_GENOME {
     TABIX_KNOWN_INDELS( known_indels.flatten().map{ it -> [[id:it.baseName], it] } )
     TABIX_PON(pon.flatten().map{ it -> [[id:it.baseName], it] })
 
+    // prepare a reference for tumor_only mode based on target_baits
+    CNVKIT_ANTITARGET(intervals_bed_combined.flatten().map{ it -> [[id:it[0].baseName], it] })
+    CNVKIT_REFERENCE(fasta, intervals_bed_combined, CNVKIT_ANTITARGET.out.bed.map{ meta, bed -> [bed]} )
+
     // prepare ascat reference files
     allele_files = ascat_alleles
     if (params.ascat_alleles && params.ascat_alleles.endsWith('.zip')) {
@@ -99,6 +106,8 @@ workflow PREPARE_GENOME {
     ch_versions = ch_versions.mix(SAMTOOLS_FAIDX.out.versions)
     ch_versions = ch_versions.mix(BWAMEM1_INDEX.out.versions)
     ch_versions = ch_versions.mix(BWAMEM2_INDEX.out.versions)
+    ch_versions = ch_versions.mix(CNVKIT_ANTITARGET.out.versions)
+    ch_versions = ch_versions.mix(CNVKIT_REFERENCE.out.versions)
     ch_versions = ch_versions.mix(GATK4_CREATESEQUENCEDICTIONARY.out.versions)
     ch_versions = ch_versions.mix(MSISENSORPRO_SCAN.out.versions)
     ch_versions = ch_versions.mix(TABIX_DBSNP.out.versions)
@@ -118,6 +127,7 @@ workflow PREPARE_GENOME {
         msisensorpro_scan                = MSISENSORPRO_SCAN.out.list.map{ meta, list -> [list] }              // path: genome_msi.list
         pon_tbi                          = TABIX_PON.out.tbi.map{ meta, tbi -> [tbi] }.collect()               // path: pon.vcf.gz.tbi
         chr_files                        = chr_files
+        cnvkit_reference                 = CNVKIT_REFERENCE.out.cnn
         allele_files                     = allele_files
         loci_files                       = loci_files
         gc_file                          = gc_file

diff --git a/subworkflows/local/tumor_variant_calling.nf b/subworkflows/local/tumor_variant_calling.nf
@@ -8,7 +8,7 @@ include { GATK_TUMOR_ONLY_SOMATIC_VARIANT_CALLING } from '../../subworkflows/nf-
 include { RUN_MANTA_TUMORONLY                     } from '../nf-core/variantcalling/manta/tumoronly/main.nf'
 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_CNVKIT                              } from '../nf-core/variantcalling/cnvkit/main.nf'
 include { RUN_MPILEUP                             } from '../nf-core/variantcalling/mpileup/main'
 include { RUN_TIDDIT                              } from '../nf-core/variantcalling/tiddit/main.nf'
 
@@ -18,6 +18,7 @@ workflow TUMOR_ONLY_VARIANT_CALLING {
         cram_recalibrated             // channel: [mandatory] cram
         bwa                           // channel: [optional] bwa
         chr_files
+        cnvkit_reference
         dbsnp                         // channel: [mandatory] dbsnp
         dbsnp_tbi                     // channel: [mandatory] dbsnp_tbi
         dict                          // channel: [mandatory] dict
@@ -99,13 +100,13 @@ workflow TUMOR_ONLY_VARIANT_CALLING {
                 [meta, cram, []]
             }
 
-        RUN_CNVKIT_TUMORONLY (  cram_recalibrated_cnvkit_tumoronly,
-                                fasta,
-                                fasta_fai,
-                                intervals_bed_combined,
-                                [] )
+        RUN_CNVKIT ( cram_recalibrated_cnvkit_tumoronly,
+                        fasta,
+                        fasta_fai,
+                        [],
+                        cnvkit_reference )
 
-        ch_versions = ch_versions.mix(RUN_CNVKIT_TUMORONLY.out.versions)
+        ch_versions = ch_versions.mix(RUN_CNVKIT.out.versions)
     }
 
     if (tools.split(',').contains('freebayes')){

diff --git a/...re/variantcalling/cnvkit/germline/main.nf → ...ows/nf-core/variantcalling/cnvkit/main.nf b/...re/variantcalling/cnvkit/germline/main.nf → ...ows/nf-core/variantcalling/cnvkit/main.nf
diff --git a/subworkflows/nf-core/variantcalling/cnvkit/somatic/main.nf b/subworkflows/nf-core/variantcalling/cnvkit/somatic/main.nf
diff --git a/subworkflows/nf-core/variantcalling/cnvkit/tumoronly/main.nf b/subworkflows/nf-core/variantcalling/cnvkit/tumoronly/main.nf