Refactor postvariantcalling. Split out valrociraptor vs other options

CHANGELOG.md

@@ -34,6 +34,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 #### Changed
 
 - [2041](https://github.com/nf-core/sarek/pull/2041) - Back to dev
+- [2043](https://github.com/nf-core/sarek/pull/2043) - Refactor postvariantcalling and split out varlociraptor from other options
 
 #### Fixed
 

README.md

@@ -66,9 +66,8 @@ Depending on the options and samples provided, the pipeline can currently perfor
   - `Sentieon Haplotyper`
   - `Strelka`
   - `TIDDIT`
-- Post-variant calling options:
-  - `BCFtools concat` for germline vcfs
-  - _Experimental Feature_ `BCFtools norm` for all vcfs
+- Post-variant calling options, one of:
+  - `BCFtools concat` for germline vcfs and/or `BCFtools norm` for all vcfs (_experimental Feature_)
   - `Varlociraptor` for all vcfs
 - Variant filtering and annotation (`SnpEff`, `Ensembl VEP`, `BCFtools annotate`)
 - Summarise and represent QC (`MultiQC`)

assets/multiqc_config.yml

@@ -13,6 +13,7 @@ report_section_order:
     order: -1002
 
 export_plots: true
+export_plots_timeout: 660
 
 disable_version_detection: true
 

conf/modules/post_variant_calling.config

@@ -18,18 +18,15 @@ process {
 
     withName: 'GERMLINE_VCFS_CONCAT' {
         ext.args   = { "-a" }
-        ext.when   = { params.concatenate_vcfs }
         publishDir = [ enabled: false ]
     }
 
     withName: 'ADD_INFO_TO_VCF' {
-        ext.when   = { params.concatenate_vcfs || params.normalize_vcfs }
         publishDir = [ enabled: false ]
     }
 
     withName: 'GERMLINE_VCFS_CONCAT_SORT' {
         ext.prefix = { "${meta.id}.germline" }
-        ext.when   = { params.concatenate_vcfs }
         publishDir = [
             mode: params.publish_dir_mode,
                 path: { "${params.outdir}/variant_calling/concat/${meta.id}/" },
@@ -39,7 +36,6 @@ process {
 
     withName: 'VCFS_NORM_SORT' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.norm" }
-        ext.when   = { params.normalize_vcfs }
         publishDir = [
             mode: params.publish_dir_mode,
                 path: { "${params.outdir}/variant_calling/normalized/${meta.id}/" },
@@ -52,22 +48,18 @@ process {
             '--multiallelics -both', // split multiallelic sites into biallelic records and both SNPs and indels should be merged separately into two records
             '--rm-dup all'           // output only the first instance of a record which is present multiple times
         ].join(' ') }
-        ext.when   = { params.normalize_vcfs }
         publishDir = [
-            mode: params.publish_dir_mode,
-                path: { "${params.outdir}/variant_calling/normalized/${meta.id}/" }
+            enabled: false
         ]
     }
 
     withName: 'TABIX_EXT_VCF' {
         ext.prefix = { "${input.baseName}" }
-        ext.when = { params.concatenate_vcfs || params.normalize_vcfs }
         publishDir = [ enabled: false ]
     }
 
     withName: 'TABIX_GERMLINE_VCFS_CONCAT_SORT' {
         ext.prefix = { "${meta.id}.germline" }
-        ext.when   = { params.concatenate_vcfs }
         publishDir = [
             mode: params.publish_dir_mode,
                 path: { "${params.outdir}/variant_calling/concat/${meta.id}/" },
@@ -77,7 +69,6 @@ process {
 
     withName: 'TABIX_VCFS_NORM_SORT' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.norm" }
-        ext.when   = { params.normalize_vcfs }
         publishDir = [
             mode: params.publish_dir_mode,
                 path: { "${params.outdir}/variant_calling/normalized/${meta.id}/" },

conf/modules/varlociraptor.config

@@ -38,7 +38,7 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_GERMLINE:RBT_VCFSPLIT|NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_TUMOR_ONLY:RBT_VCFSPLIT' {
+    withName: '.*:VCF_VARLOCIRAPTOR_GERMLINE:RBT_VCFSPLIT|.*:VCF_VARLOCIRAPTOR_TUMOR_ONLY:RBT_VCFSPLIT' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.split" }
         publishDir = [
             enabled: false
@@ -52,22 +52,31 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_GERMLINE:VARLOCIRAPTOR_CALLVARIANTS|NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_TUMOR_ONLY:VARLOCIRAPTOR_CALLVARIANTS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_GERMLINE:VARLOCIRAPTOR_CALLVARIANTS|.*:VCF_VARLOCIRAPTOR_TUMOR_ONLY:VARLOCIRAPTOR_CALLVARIANTS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.call.${meta.chunk}" }
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_GERMLINE:SORT_CALLED_CHUNKS|NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_TUMOR_ONLY:SORT_CALLED_CHUNKS' {
-        ext.prefix = { "${meta.id}.${meta.variantcaller}.sort.${meta.chunk}" }
+    withName: 'SORT_CALLED_CHUNKS' {
         ext.args   = { '--output-type z --write-index=tbi' }
         publishDir = [
-            enabled: false
+            mode: params.publish_dir_mode,
+            path: { "${params.outdir}/variant_calling/" },
+            saveAs: { params.varlociraptor_chunk_size > 1 ? null : "varlociraptor/${meta.id}/${it}" }
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_GERMLINE:CONCAT_CALLED_CHUNKS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_GERMLINE:SORT_CALLED_CHUNKS' {
+        ext.prefix = { params.varlociraptor_chunk_size > 1 ? "${meta.id}.${meta.variantcaller}.sort.${meta.chunk}" : "${meta.id}.${meta.variantcaller}.germline.varlociraptor"}
+    }
+
+    withName: '.*:VCF_VARLOCIRAPTOR_TUMOR_ONLY:SORT_CALLED_CHUNKS' {
+        ext.prefix = { params.varlociraptor_chunk_size > 1 ? "${meta.id}.${meta.variantcaller}.sort.${meta.chunk}" : "${meta.id}.${meta.variantcaller}.tumor_only.varlociraptor"}
+    }
+
+    withName: '.*:VCF_VARLOCIRAPTOR_GERMLINE:CONCAT_CALLED_CHUNKS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.germline.varlociraptor" }
         ext.args   = { '--allow-overlaps --output-type z --write-index=tbi' }
         publishDir = [
@@ -77,7 +86,7 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_TUMOR_ONLY:CONCAT_CALLED_CHUNKS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_TUMOR_ONLY:CONCAT_CALLED_CHUNKS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.tumor_only.varlociraptor" }
         ext.args   = { '--allow-overlaps --output-type z --write-index=tbi' }
         publishDir = [
@@ -90,7 +99,7 @@ process {
     //
     // VCF_VARLOCIRAPTOR_SOMATIC MODULES
     //
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:ALIGNMENTPROPERTIES_NORMAL' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:ALIGNMENTPROPERTIES_NORMAL' {
         ext.prefix = { "${meta.normal_id}.normal" }
         publishDir = [
             mode: params.publish_dir_mode,
@@ -99,13 +108,13 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:TABIX_GERMLINE|NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:TABIX_SOMATIC' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:TABIX_GERMLINE|.*:VCF_VARLOCIRAPTOR_SOMATIC:TABIX_SOMATIC' {
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:CONCAT_SOMATIC_STRELKA' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:CONCAT_SOMATIC_STRELKA' {
         ext.prefix = { "${meta.id}.strelka" }
         ext.args   = { '--allow-overlaps --output-type z --write-index=tbi' }
         publishDir = [
@@ -115,7 +124,7 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:MERGE_GERMLINE_SOMATIC_VCFS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:MERGE_GERMLINE_SOMATIC_VCFS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.merged" }
         ext.args   = { '--output-type z' }
         publishDir = [
@@ -125,21 +134,21 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:RBT_VCFSPLIT' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:RBT_VCFSPLIT' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.split" }
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:PREPROCESS_NORMAL' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:PREPROCESS_NORMAL' {
         ext.prefix = { "${meta.normal_id}.${meta.variantcaller}.preprocess.${meta.chunk}" }
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:ALIGNMENTPROPERTIES_TUMOR' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:ALIGNMENTPROPERTIES_TUMOR' {
         ext.prefix = { "${meta.tumor_id}.tumor" }
         publishDir = [
             mode: params.publish_dir_mode,
@@ -148,29 +157,25 @@ process {
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:PREPROCESS_TUMOR' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:PREPROCESS_TUMOR' {
         ext.prefix = { "${meta.tumor_id}.${meta.variantcaller}.preprocess.${meta.chunk}" }
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:VARLOCIRAPTOR_CALLVARIANTS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:VARLOCIRAPTOR_CALLVARIANTS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.call.${meta.chunk}" }
         publishDir = [
             enabled: false
         ]
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:SORT_CALLED_CHUNKS' {
-        ext.prefix = { "${meta.id}.${meta.variantcaller}.sort.${meta.chunk}" }
-        ext.args   = { '--output-type z --write-index=tbi' }
-        publishDir = [
-            enabled: false
-        ]
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:SORT_CALLED_CHUNKS' {
+        ext.prefix = { params.varlociraptor_chunk_size > 1 ? "${meta.id}.${meta.variantcaller}.sort.${meta.chunk}" : "${meta.id}.${meta.variantcaller}.somatic.varlociraptor"}
     }
 
-    withName: 'NFCORE_SAREK:SAREK:POST_VARIANTCALLING:VCF_VARLOCIRAPTOR_SOMATIC:CONCAT_CALLED_CHUNKS' {
+    withName: '.*:VCF_VARLOCIRAPTOR_SOMATIC:CONCAT_CALLED_CHUNKS' {
         ext.prefix = { "${meta.id}.${meta.variantcaller}.somatic.varlociraptor" }
         ext.args   = { '--allow-overlaps --output-type z --write-index=tbi' }
         publishDir = [

conf/test.config

@@ -13,7 +13,7 @@ process {
     resourceLimits = [
         cpus: 4,
         memory: '15.GB',
-        time: '1.h'
+        time: '2.h'
     ]
     withName:BWAMEM2_INDEX {
         memory = { 6.GB }

main.nf

@@ -266,6 +266,10 @@ workflow NFCORE_SAREK {
         PREPARE_GENOME.out.pon,
         PREPARE_GENOME.out.pon_tbi,
         params.sentieon_dnascope_model ? Channel.fromPath(params.sentieon_dnascope_model).collect() : Channel.value([]),
+         // Set Varlociraptor reference files
+        params.varlociraptor_scenario_germline   ? Channel.fromPath(params.varlociraptor_scenario_germline).map { it -> [[id: it.baseName - '.yte'], it] }.collect()    : Channel.fromPath("${projectDir}/assets/varlociraptor_germline.yte.yaml").collect(),
+        params.varlociraptor_scenario_somatic    ? Channel.fromPath(params.varlociraptor_scenario_somatic).map { it -> [[id: it.baseName - '.yte'], it] }.collect()     : Channel.fromPath("${projectDir}/assets/varlociraptor_somatic.yte.yaml").collect(),
+        params.varlociraptor_scenario_tumor_only ? Channel.fromPath(params.varlociraptor_scenario_tumor_only).map { it -> [[id: it.baseName - '.yte'], it] }.collect()  : Channel.fromPath("${projectDir}/assets/varlociraptor_tumor_only.yte.yaml").collect(),
         snpeff_cache,
         params.snpeff_db,
         vep_cache,

nextflow_schema.json

@@ -516,7 +516,8 @@
                 "varlociraptor_chunk_size": {
                     "type": "integer",
                     "default": 15,
-                    "description": "Number of chunks to split the vcf-files for varlociraptor",
+                    "minimum": 1,
+                    "description": "Number of chunks to split the vcf-files for varlociraptor. Minimum 1, indicates no splitting",
                     "hidden": true
                 },
                 "varlociraptor_scenario_germline": {