Fix unique lanes for freebayes, groupKey, add new indices, copy meta maps

CHANGELOG.md

@@ -73,6 +73,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 - [#513](https://github.com/nf-core/sarek/pull/513), [#527](https://github.com/nf-core/sarek/pull/527) - CNV is back
 - [#529](https://github.com/nf-core/sarek/pull/529) - Do not save `versions.yml` files
 - [#524](https://github.com/nf-core/sarek/pull/524) - Fix intervals usage by counting the actual list of scatter/gather files produced and not overall number of intervals
+- [#549](https://github.com/nf-core/sarek/pull/549) - Fix unique lanes required for Freebayes: issue [#311](https://github.com/nf-core/sarek/issues/311)
 
 ### Deprecated
 

conf/modules.config

@@ -475,13 +475,9 @@ process{
         // For unknown reasons, CONCAT_VCF sometimes fails with SIGPIPE
         // (exit code 141). Rerunning the process will usually work.
         errorStrategy   = {task.exitStatus == 141 ? 'retry' : 'terminate'}
-        ext.args         = { params.no_intervals ?  "-n" : "" }
-    }
-    withName : 'TABIX_VC_.*' {
-        ext.when         = { params.no_intervals }
+        ext.args         = { params.no_intervals ?  "-n" : "" } //Why ConcatVCF is never run when no_intervals is set..
     }
     withName : 'BGZIP_VC_.*' {
-        ext.when         = { !params.no_intervals }
         publishDir       = [
             enabled: false
         ]
@@ -525,15 +521,23 @@ process{
         ]
     }
     withName: 'FREEBAYES' {
+        ext.prefix       = {"${meta.id}.freebayes"} //To make sure no naming conflicts ensue with module BCFTOOLS_SORT & the naming being correct in the output folder
         ext.args         = '--min-alternate-fraction 0.1 --min-mapping-quality 1'
         ext.when         = { params.tools && params.tools.contains('freebayes') }
+        publishDir       = [
+            enabled: false
+        ]
+    }
+
+    withName: 'BCFTOOLS_SORT' {
         publishDir       = [
             mode: params.publish_dir_mode,
             path: { "${params.outdir}/variant_calling/${meta.id}/freebayes" },
             pattern: "*vcf.gz",
             saveAs: { meta.num_intervals > 1 ? null : it }
         ]
     }
+
     withName : 'TABIX_VC_FREEBAYES' {
         publishDir       = [
             mode: params.publish_dir_mode,
@@ -900,6 +904,18 @@ process{
             }
     }
 
+    //FREEBAYES
+    withName: 'NFCORE_SAREK:SAREK:PAIR_VARIANT_CALLING:RUN_FREEBAYES_SOMATIC:FREEBAYES' {
+        ext.args = "--pooled-continuous \
+                    --pooled-discrete \
+                    --genotype-qualities \
+                    --report-genotype-likelihood-max \
+                    --allele-balance-priors-off \
+                    --min-alternate-fraction 0.03 \
+                    --min-repeat-entropy 1 \
+                    --min-alternate-count 2 "
+    }
+
     //MANTA
     withName: 'CONCAT_MANTA_SOMATIC' {
         ext.prefix       = {"${meta.id}.somatic_sv"}

modules.json

@@ -6,6 +6,9 @@
             "ascat": {
                 "git_sha": "f0800157544a82ae222931764483331a81812012"
             },
+            "bcftools/sort": {
+                "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"
+            },
             "bcftools/stats": {
                 "git_sha": "e745e167c1020928ef20ea1397b6b4d230681b4d"
             },
@@ -130,7 +133,7 @@
                 "git_sha": "169b2b96c1167f89ab07127b7057c1d90a6996c7"
             },
             "gatk4/markduplicates": {
-                "git_sha": "169b2b96c1167f89ab07127b7057c1d90a6996c7"
+                "git_sha": "df2620cfc7e4c21b14ed03c1c928f09fbabf83c4"
             },
             "gatk4/markduplicatesspark": {
                 "git_sha": "e04970b7d249365cafa5a52912f9a28840481c05"

subworkflows/local/germline_variant_calling.nf

@@ -38,32 +38,28 @@ workflow GERMLINE_VARIANT_CALLING {
     // Remap channel with intervals
     cram_recalibrated_intervals = cram_recalibrated.combine(intervals)
         .map{ meta, cram, crai, intervals, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + intervals.baseName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + intervals.baseName
 
             //If no interval file provided (0) then add empty list
             intervals_new = num_intervals == 0 ? [] : intervals
 
-            [new_meta, cram, crai, intervals_new]
+            [[patient:meta.patient, sample:meta.sample, gender:meta.gender, status:meta.status, id:new_id, data_type:meta.data_type, num_intervals:num_intervals],
+            cram, crai, intervals_new]
         }
 
     // Remap channel with gzipped intervals + indexes
     cram_recalibrated_intervals_gz_tbi = cram_recalibrated.combine(intervals_bed_gz_tbi)
         .map{ meta, cram, crai, bed_tbi, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + bed_tbi[0].simpleName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + bed_tbi[0].simpleName
 
             //If no interval file provided (0) then add empty list
             bed_new = num_intervals == 0 ? [] : bed_tbi[0]
             tbi_new = num_intervals == 0 ? [] : bed_tbi[1]
 
-            [new_meta, cram, crai, bed_new, tbi_new]
+            [[patient:meta.patient, sample:meta.sample, gender:meta.gender, status:meta.status, id:new_id, data_type:meta.data_type, num_intervals:num_intervals],
+            cram, crai, bed_new, tbi_new]
         }
 
     // DEEPVARIANT

subworkflows/local/pair_variant_calling.nf

@@ -4,6 +4,7 @@
 include { GATK_TUMOR_NORMAL_SOMATIC_VARIANT_CALLING } from '../../subworkflows/nf-core/gatk4/tumor_normal_somatic_variant_calling/main'
 include { MSISENSORPRO_MSI_SOMATIC                  } from '../../modules/nf-core/modules/msisensorpro/msi_somatic/main'
 include { RUN_CONTROLFREEC_SOMATIC                  } from '../nf-core/variantcalling/controlfreec/somatic/main.nf'
+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'
 
@@ -34,6 +35,7 @@ workflow PAIR_VARIANT_CALLING {
     ch_versions          = Channel.empty()
 
     //TODO: Temporary until the if's can be removed and printing to terminal is prevented with "when" in the modules.config
+    freebayes_vcf        = Channel.empty()
     manta_vcf            = Channel.empty()
     strelka_vcf          = Channel.empty()
     msisensorpro_output  = Channel.empty()
@@ -42,32 +44,28 @@ workflow PAIR_VARIANT_CALLING {
     // Remap channel with intervals
     cram_pair_intervals = cram_pair.combine(intervals)
         .map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai, intervals, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + intervals.baseName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + intervals.baseName
 
             //If no interval file provided (0) then add empty list
             intervals_new = num_intervals == 0 ? [] : intervals
 
-            [new_meta, normal_cram, normal_crai, tumor_cram, tumor_crai, intervals_new]
+            [[patient:meta.patient, normal_id:meta.normal_id, tumor_id:meta.tumor_id, gender:meta.gender, id:new_id, num_intervals:num_intervals],
+            normal_cram, normal_crai, tumor_cram, tumor_crai, intervals_new]
         }
 
     // Remap channel with gzipped intervals + indexes
     cram_pair_intervals_gz_tbi = cram_pair.combine(intervals_bed_gz_tbi)
         .map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai, bed_tbi, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + bed_tbi[0].simpleName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + bed_tbi[0].simpleName
 
             //If no interval file provided (0) then add empty list
             bed_new = num_intervals == 0 ? [] : bed_tbi[0]
             tbi_new = num_intervals == 0 ? [] : bed_tbi[1]
 
-            [new_meta, normal_cram, normal_crai, tumor_cram, tumor_crai, bed_new, tbi_new]
+            [[patient:meta.patient, normal_id:meta.normal_id, tumor_id:meta.tumor_id, gender:meta.gender, id:new_id, num_intervals:num_intervals],
+            normal_cram, normal_crai, tumor_cram, tumor_crai, bed_new, tbi_new]
         }
 
     if (tools.contains('controlfreec')){
@@ -93,6 +91,13 @@ workflow PAIR_VARIANT_CALLING {
         ch_versions = ch_versions.mix(RUN_CONTROLFREEC_SOMATIC.out.versions)
     }
 
+    if (tools.contains('freebayes')){
+        RUN_FREEBAYES_SOMATIC(cram_pair_intervals, fasta, fasta_fai, intervals_bed_combine_gz)
+
+        freebayes_vcf = RUN_FREEBAYES_SOMATIC.out.freebayes_vcf
+        ch_versions   = ch_versions.mix(RUN_FREEBAYES_SOMATIC.out.versions)
+    }
+
     if (tools.contains('manta')) {
         RUN_MANTA_SOMATIC(  cram_pair_intervals_gz_tbi,
                             fasta,
@@ -114,14 +119,14 @@ workflow PAIR_VARIANT_CALLING {
                                         .map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai, vcf, bed_tbi, num_intervals ->
 
                                              // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-                                            new_meta.id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + bed_tbi[0].simpleName
-                                            new_meta.num_intervals = num_intervals
+                                            new_id = num_intervals <= 1 ? meta.tumor_id + "_vs_" + meta.normal_id : meta.tumor_id + "_vs_" + meta.normal_id + "_" + bed_tbi[0].simpleName
 
                                             //If no interval file provided (0) then add empty list
                                             bed_new = num_intervals == 0 ? [] : bed_tbi[0]
                                             tbi_new = num_intervals == 0 ? [] : bed_tbi[1]
 
-                                            [new_meta, normal_cram, normal_crai, tumor_cram, tumor_crai, vcf, vcf_tbi, bed_new, tbi_new]
+                                            [[patient:meta.patient, normal_id:meta.normal_id, tumor_id:meta.tumor_id, gender:meta.gender, id:new_id, num_intervals:num_intervals],
+                                            normal_cram, normal_crai, tumor_cram, tumor_crai, vcf, vcf_tbi, bed_new, tbi_new]
                                         }
         } else {
             cram_pair_strelka = cram_pair_intervals_gz_tbi.map{
@@ -172,6 +177,7 @@ workflow PAIR_VARIANT_CALLING {
     // }
 
     emit:
+    freebayes_vcf
     manta_vcf
     msisensorpro_output
     mutect2_vcf

subworkflows/local/tumor_variant_calling.nf

@@ -3,7 +3,6 @@
 // Should be only run on patients without normal sample
 //
 
-//include { RUN_CONTROLFREEC                        } from '../nf-core/variantcalling/controlfreec/main.nf'
 include { RUN_FREEBAYES                           } from '../nf-core/variantcalling/freebayes/main.nf'
 include { GATK_TUMOR_ONLY_SOMATIC_VARIANT_CALLING } from '../../subworkflows/nf-core/gatk4/tumor_only_somatic_variant_calling/main'
 include { RUN_MANTA_TUMORONLY                     } from '../nf-core/variantcalling/manta/tumoronly/main.nf'
@@ -44,32 +43,28 @@ workflow TUMOR_ONLY_VARIANT_CALLING {
     // Remap channel with intervals
     cram_recalibrated_intervals = cram_recalibrated.combine(intervals)
         .map{ meta, cram, crai, intervals, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + intervals.baseName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + intervals.baseName
 
             //If no interval file provided (0) then add empty list
             intervals_new = num_intervals == 0 ? [] : intervals
 
-            [new_meta, cram, crai, intervals_new]
+            [[patient:meta.patient, sample:meta.sample, gender:meta.gender, status:meta.status, id:new_id, data_type:meta.data_type, num_intervals:num_intervals],
+            cram, crai, intervals_new]
         }
 
     // Remap channel with gzipped intervals + indexes
     cram_recalibrated_intervals_gz_tbi = cram_recalibrated.combine(intervals_bed_gz_tbi)
         .map{ meta, cram, crai, bed_tbi, num_intervals ->
-            new_meta = meta.clone()
-
             // If either no scatter/gather is done, i.e. no interval (0) or one interval (1), then don't rename samples
-            new_meta.id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + bed_tbi[0].simpleName
-            new_meta.num_intervals = num_intervals
+            new_id = num_intervals <= 1 ? meta.sample : meta.sample + "_" + bed_tbi[0].simpleName
 
             //If no interval file provided (0) then add empty list
             bed_new = num_intervals == 0 ? [] : bed_tbi[0]
             tbi_new = num_intervals == 0 ? [] : bed_tbi[1]
 
-            [new_meta, cram, crai, bed_new, tbi_new]
+            [[patient:meta.patient, sample:meta.sample, gender:meta.gender, status:meta.status, id:new_id, data_type:meta.data_type, num_intervals:num_intervals],
+            cram, crai, bed_new, tbi_new]
         }
 
     if(tools.contains('controlfreec')){