Merge pull request #552 from nf-core/mito_wes

Skip mitochondrial analysis for WES
nf-core · May 20, 2024 · 6521bba · 6521bba
2 parents 4a9bb55 + 8c93dc5
commit 6521bba
Show file tree

Hide file tree

Showing 13 changed files with 144 additions and 131 deletions.
diff --git a/CHANGELOG.md b/CHANGELOG.md
@@ -7,6 +7,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### `Added`
 
+- A new parameter `run_mt_for_wes` to turn on mitochondrial analysis for targeted analysis [#552](https://github.com/nf-core/raredisease/pull/552)
 - A new parameter `bwa_as_fallback` to switch aligner to bwa in case bwamem2 fails [#551](https://github.com/nf-core/raredisease/pull/551)
 
 ### `Changed`
@@ -22,8 +23,10 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Parameters
 
-| Old parameter | New parameter |
-| ------------- | ------------- |
+| Old parameter | New parameter   |
+| ------------- | --------------- |
+|               | bwa_as_fallback |
+|               | run_mt_for_wes  |
 
 :::note
 Parameter has been updated if both old and new parameter information is present.

diff --git a/conf/modules/align_MT.config b/conf/modules/align_MT.config
@@ -19,19 +19,16 @@ process {
 
     withName: '.*ALIGN_MT:BWAMEM2_MEM_MT' {
         ext.args = { "-M -K 100000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes")) }
         ext.prefix = { "${meta.id}_sorted" }
     }
 
     withName: '.*ALIGN_MT:BWA_MEM_MT' {
         ext.args = { "-M -K 100000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes")) }
         ext.prefix = { "${meta.id}_sorted" }
     }
 
     withName: '.*ALIGN_MT:SENTIEON_BWAMEM_MT' {
         ext.args = { "-M -K 10000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes"))}
         ext.prefix = { "${meta.id}_sorted" }
     }
 
@@ -70,19 +67,16 @@ process {
 
     withName: '.*ALIGN_MT_SHIFT:BWAMEM2_MEM_MT' {
         ext.args = { "-M -K 100000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes")) }
         ext.prefix = { "${meta.id}_sorted_shifted" }
     }
 
     withName: '.*ALIGN_MT_SHIFT:SENTIEON_BWAMEM_MT' {
         ext.args = { "-M -K 10000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes"))}
         ext.prefix = { "${meta.id}_sorted_shifted" }
     }
 
     withName: '.*ALIGN_MT_SHIFT:BWA_MEM_MT' {
         ext.args = { "-M -K 100000000 -R ${meta.read_group}" }
-        ext.when = { !(params.analysis_type.equals("wes")) }
         ext.prefix = { "${meta.id}_sorted_shifted" }
     }
 

diff --git a/conf/modules/call_repeat_expansions.config b/conf/modules/call_repeat_expansions.config
@@ -16,10 +16,6 @@
 //
 
 process {
-    withName: '.*CALL_REPEAT_EXPANSIONS:.*' {
-        ext.when = { params.analysis_type != "wes" }
-    }
-
     withName: '.*CALL_REPEAT_EXPANSIONS:EXPANSIONHUNTER' {
         ext.args =  { ("${meta.sex}" == '1') ? '--sex male' : '--sex female' }
         ext.prefix = { "${meta.id}_exphunter" }

diff --git a/conf/modules/prepare_references.config b/conf/modules/prepare_references.config
@@ -30,19 +30,19 @@ process {
     }
 
     withName: '.*PREPARE_REFERENCES:BWAMEM2_INDEX_MT_SHIFT' {
-        ext.when = { !(params.analysis_type == "wes") && params.aligner == "bwamem2"}
+        ext.when = { (params.analysis_type.equals("wgs") || params.run_mt_for_wes) && params.aligner == "bwamem2"}
     }
 
     withName: '.*PREPARE_REFERENCES:SENTIEON_BWAINDEX_GENOME' {
         ext.when = {!params.bwa && params.aligner == "sentieon"}
     }
 
     withName: '.*PREPARE_REFERENCES:SENTIEON_BWAINDEX_MT_SHIFT' {
-        ext.when = { !(params.analysis_type == "wes") && params.aligner == "sentieon"}
+        ext.when = { (params.analysis_type.equals("wgs") || params.run_mt_for_wes) && params.aligner == "sentieon"}
     }
 
     withName: '.*PREPARE_REFERENCES:BWA_INDEX_MT_SHIFT' {
-        ext.when = { !(params.analysis_type == "wes") && params.aligner == "bwa"}
+        ext.when = { (params.analysis_type.equals("wgs") || params.run_mt_for_wes) && params.aligner == "bwa"}
     }
 
     withName: '.*PREPARE_REFERENCES:BWA_INDEX_GENOME' {
@@ -63,7 +63,7 @@ process {
     }
 
     withName: '.*PREPARE_REFERENCES:SAMTOOLS_FAIDX_MT_SHIFT' {
-        ext.when = { !(params.analysis_type == "wes")}
+        ext.when = { (params.analysis_type.equals("wgs") || params.run_mt_for_wes) }
     }
 
     withName: '.*PREPARE_REFERENCES:GATK_SD' {
@@ -75,7 +75,7 @@ process {
     }
 
     withName: '.*PREPARE_REFERENCES:GATK_SD_MT_SHIFT' {
-        ext.when = { !(params.analysis_type == "wes")}
+        ext.when = { (params.analysis_type.equals("wgs") || params.run_mt_for_wes)}
     }
 
     withName: '.*PREPARE_REFERENCES:TABIX_DBSNP' {

diff --git a/conf/modules/qc_bam.config b/conf/modules/qc_bam.config
@@ -74,23 +74,23 @@ process {
 
     withName: '.*QC_BAM:PICARD_COLLECTWGSMETRICS' {
         ext.args = "--TMP_DIR ."
-        ext.when = { params.analysis_type.equals("wgs") && !params.aligner.equals("sentieon") }
+        ext.when = { !params.aligner.equals("sentieon") }
         ext.prefix = { "${meta.id}_wgsmetrics" }
     }
 
     withName: '.*QC_BAM:PICARD_COLLECTWGSMETRICS_Y' {
         ext.args = "--TMP_DIR ."
-        ext.when = { params.analysis_type.equals("wgs") && !params.aligner.equals("sentieon") }
+        ext.when = { !params.aligner.equals("sentieon") }
         ext.prefix = { "${meta.id}_wgsmetrics_y" }
     }
 
     withName: '.*QC_BAM:SENTIEON_WGSMETRICS' {
-        ext.when = { params.analysis_type.equals("wgs") && params.aligner.equals("sentieon") }
+        ext.when = { params.aligner.equals("sentieon") }
         ext.prefix = { "${meta.id}_wgsmetrics" }
     }
 
     withName: '.*QC_BAM:SENTIEON_WGSMETRICS_Y' {
-        ext.when = { params.analysis_type.equals("wgs") && params.aligner.equals("sentieon") }
+        ext.when = { params.aligner.equals("sentieon") }
         ext.prefix = { "${meta.id}_wgsmetrics_y" }
     }
 }
diff --git a/nextflow.config b/nextflow.config
@@ -24,6 +24,7 @@ params {
     analysis_type              = 'wgs'
     bwa_as_fallback            = false
     bait_padding               = 100
+    run_mt_for_wes             = false
     run_rtgvcfeval             = false
     save_mapped_as_cram        = false
     skip_eklipse               = false

diff --git a/nextflow_schema.json b/nextflow_schema.json
@@ -485,6 +485,11 @@
                     "fa_icon": "fas fa-align-center",
                     "enum": ["xy", "hetx", "sry"]
                 },
+                "run_mt_for_wes": {
+                    "type": "boolean",
+                    "description": "Specifies whether to run mitochondrial analysis for wes samples",
+                    "fa_icon": "fas fa-toggle-on"
+                },
                 "run_rtgvcfeval": {
                     "type": "boolean",
                     "description": "Specifies whether to run rtgtools' vcfeval",

diff --git a/subworkflows/local/align.nf b/subworkflows/local/align.nf
@@ -26,12 +26,18 @@ workflow ALIGN {
         val_platform             // string:  [mandatory] illumina or a different technology
 
     main:
-        ch_versions       = Channel.empty()
-        ch_fastp_json     = Channel.empty()
-        ch_bwamem2_bam    = Channel.empty()
-        ch_sentieon_bam   = Channel.empty()
-        ch_bwamem2_bai    = Channel.empty()
-        ch_sentieon_bai   = Channel.empty()
+        ch_bwamem2_bam        = Channel.empty()
+        ch_bwamem2_bai        = Channel.empty()
+        ch_fastp_json         = Channel.empty()
+        ch_mt_bam_bai         = Channel.empty()
+        ch_mt_marked_bam      = Channel.empty()
+        ch_mt_marked_bai      = Channel.empty()
+        ch_mtshift_bam_bai    = Channel.empty()
+        ch_mtshift_marked_bam = Channel.empty()
+        ch_mtshift_marked_bai = Channel.empty()
+        ch_sentieon_bam       = Channel.empty()
+        ch_sentieon_bai       = Channel.empty()
+        ch_versions           = Channel.empty()
 
         if (!params.skip_fastp) {
             FASTP (ch_reads, [], false, false)
@@ -70,48 +76,49 @@ workflow ALIGN {
         ch_genome_bam_bai    = ch_genome_marked_bam.join(ch_genome_marked_bai, failOnMismatch:true, failOnDuplicate:true)
 
         // PREPARING READS FOR MT ALIGNMENT
-        CONVERT_MT_BAM_TO_FASTQ (
-            ch_genome_bam_bai,
-            ch_genome_fasta,
-            ch_genome_fai,
-            ch_genome_dictionary
-        )
-
-        ALIGN_MT (
-            CONVERT_MT_BAM_TO_FASTQ.out.fastq,
-            CONVERT_MT_BAM_TO_FASTQ.out.bam,
-            ch_genome_bwaindex,
-            ch_genome_bwamem2index,
-            ch_genome_fasta,
-            ch_genome_dictionary,
-            ch_genome_fai
-        )
+        if (params.analysis_type.equals("wgs") || params.run_mt_for_wes) {
+            CONVERT_MT_BAM_TO_FASTQ (
+                ch_genome_bam_bai,
+                ch_genome_fasta,
+                ch_genome_fai,
+                ch_genome_dictionary
+            )
 
-        ALIGN_MT_SHIFT (
-            CONVERT_MT_BAM_TO_FASTQ.out.fastq,
-            CONVERT_MT_BAM_TO_FASTQ.out.bam,
-            ch_mtshift_bwaindex,
-            ch_mtshift_bwamem2index,
-            ch_mtshift_fasta,
-            ch_mtshift_dictionary,
-            ch_mtshift_fai
-        )
+            ALIGN_MT (
+                CONVERT_MT_BAM_TO_FASTQ.out.fastq,
+                CONVERT_MT_BAM_TO_FASTQ.out.bam,
+                ch_genome_bwaindex,
+                ch_genome_bwamem2index,
+                ch_genome_fasta,
+                ch_genome_dictionary,
+                ch_genome_fai
+            )
 
-        ch_mt_marked_bam = ALIGN_MT.out.marked_bam
-        ch_mt_marked_bai = ALIGN_MT.out.marked_bai
-        ch_mt_bam_bai    = ch_mt_marked_bam.join(ch_mt_marked_bai, failOnMismatch:true, failOnDuplicate:true)
+            ALIGN_MT_SHIFT (
+                CONVERT_MT_BAM_TO_FASTQ.out.fastq,
+                CONVERT_MT_BAM_TO_FASTQ.out.bam,
+                ch_mtshift_bwaindex,
+                ch_mtshift_bwamem2index,
+                ch_mtshift_fasta,
+                ch_mtshift_dictionary,
+                ch_mtshift_fai
+            )
 
-        ch_mtshift_marked_bam = ALIGN_MT_SHIFT.out.marked_bam
-        ch_mtshift_marked_bai = ALIGN_MT_SHIFT.out.marked_bai
-        ch_mtshift_bam_bai    = ch_mtshift_marked_bam.join(ch_mtshift_marked_bai, failOnMismatch:true, failOnDuplicate:true)
+            ch_mt_marked_bam      = ALIGN_MT.out.marked_bam
+            ch_mt_marked_bai      = ALIGN_MT.out.marked_bai
+            ch_mt_bam_bai         = ch_mt_marked_bam.join(ch_mt_marked_bai, failOnMismatch:true, failOnDuplicate:true)
+            ch_mtshift_marked_bam = ALIGN_MT_SHIFT.out.marked_bam
+            ch_mtshift_marked_bai = ALIGN_MT_SHIFT.out.marked_bai
+            ch_mtshift_bam_bai    = ch_mtshift_marked_bam.join(ch_mtshift_marked_bai, failOnMismatch:true, failOnDuplicate:true)
+            ch_versions           = ch_versions.mix(ALIGN_MT.out.versions,
+                                        ALIGN_MT_SHIFT.out.versions,
+                                        CONVERT_MT_BAM_TO_FASTQ.out.versions)
+        }
 
         if (params.save_mapped_as_cram) {
             SAMTOOLS_VIEW( ch_genome_bam_bai, ch_genome_fasta, [] )
             ch_versions   = ch_versions.mix(SAMTOOLS_VIEW.out.versions)
         }
-        ch_versions   = ch_versions.mix(ALIGN_MT.out.versions,
-                                        ALIGN_MT_SHIFT.out.versions,
-                                        CONVERT_MT_BAM_TO_FASTQ.out.versions)
 
     emit:
         fastp_json         = ch_fastp_json         // channel: [ val(meta), path(json) ]

diff --git a/subworkflows/local/alignment/align_MT.nf b/subworkflows/local/alignment/align_MT.nf
@@ -23,25 +23,21 @@ workflow ALIGN_MT {
 
     main:
         ch_versions     = Channel.empty()
-        ch_bwa_bam      = Channel.empty()
-        ch_bwamem2_bam  = Channel.empty()
-        ch_sentieon_bam = Channel.empty()
 
         if (params.aligner.equals("bwamem2")) {
             BWAMEM2_MEM_MT (ch_fastq, ch_bwamem2index, true)
-            ch_bwamem2_bam = BWAMEM2_MEM_MT.out.bam
+            ch_align       = BWAMEM2_MEM_MT.out.bam
             ch_versions    = ch_versions.mix(BWAMEM2_MEM_MT.out.versions.first())
         } else if (params.aligner.equals("sentieon")) {
             SENTIEON_BWAMEM_MT ( ch_fastq, ch_bwaindex, ch_fasta, ch_fai )
-            ch_sentieon_bam = SENTIEON_BWAMEM_MT.out.bam_and_bai.map{ meta, bam, bai -> [meta, bam] }
-            ch_versions     = ch_versions.mix(SENTIEON_BWAMEM_MT.out.versions.first())
+            ch_align       = SENTIEON_BWAMEM_MT.out.bam_and_bai.map{ meta, bam, bai -> [meta, bam] }
+            ch_versions    = ch_versions.mix(SENTIEON_BWAMEM_MT.out.versions.first())
         } else if (params.aligner.equals("bwa")) {
             BWA_MEM_MT ( ch_fastq, ch_bwaindex, true )
-            ch_bwa_bam      = BWA_MEM_MT.out.bam
-            ch_versions     = ch_versions.mix(BWA_MEM_MT.out.versions.first())
+            ch_align       = BWA_MEM_MT.out.bam
+            ch_versions    = ch_versions.mix(BWA_MEM_MT.out.versions.first())
         }
-        Channel.empty()
-            .mix(ch_bwamem2_bam, ch_sentieon_bam, ch_bwa_bam)
+        ch_align
             .join(ch_ubam, failOnMismatch:true, failOnDuplicate:true)
             .set {ch_bam_ubam}
 

diff --git a/subworkflows/local/call_snv.nf b/subworkflows/local/call_snv.nf
@@ -38,6 +38,8 @@ workflow CALL_SNV {
         ch_deepvar_tbi   = Channel.empty()
         ch_deepvar_gvcf  = Channel.empty()
         ch_deepvar_gtbi  = Channel.empty()
+        ch_mt_vcf        = Channel.empty()
+        ch_mt_tabix      = Channel.empty()
         ch_sentieon_vcf  = Channel.empty()
         ch_sentieon_tbi  = Channel.empty()
         ch_sentieon_gvcf = Channel.empty()
@@ -93,46 +95,49 @@ workflow CALL_SNV {
         ch_genome_tabix     = GATK4_SELECTVARIANTS.out.tbi
         ch_genome_vcf_tabix = ch_genome_vcf.join(ch_genome_tabix, failOnMismatch:true, failOnDuplicate:true)
 
-        CALL_SNV_MT(
-            ch_mt_bam_bai,
-            ch_genome_fasta,
-            ch_genome_fai,
-            ch_genome_dictionary,
-            ch_mt_intervals
-        )
-
-        CALL_SNV_MT_SHIFT(
-            ch_mtshift_bam_bai,
-            ch_mtshift_fasta,
-            ch_mtshift_fai,
-            ch_mtshift_dictionary,
-            ch_mtshift_intervals
-        )
+        if (params.analysis_type.equals("wgs") || params.run_mt_for_wes) {
+            CALL_SNV_MT(
+                ch_mt_bam_bai,
+                ch_genome_fasta,
+                ch_genome_fai,
+                ch_genome_dictionary,
+                ch_mt_intervals
+            )
 
-        POSTPROCESS_MT_CALLS(
-            CALL_SNV_MT.out.vcf,
-            CALL_SNV_MT_SHIFT.out.vcf,
-            ch_genome_fasta,
-            ch_genome_dictionary,
-            ch_genome_fai,
-            ch_mtshift_backchain,
-            ch_case_info,
-            ch_foundin_header,
-            ch_genome_chrsizes
-        )
+            CALL_SNV_MT_SHIFT(
+                ch_mtshift_bam_bai,
+                ch_mtshift_fasta,
+                ch_mtshift_fai,
+                ch_mtshift_dictionary,
+                ch_mtshift_intervals
+            )
 
-        ch_versions = ch_versions.mix(CALL_SNV_MT.out.versions)
-        ch_versions = ch_versions.mix(CALL_SNV_MT_SHIFT.out.versions)
-        ch_versions = ch_versions.mix(POSTPROCESS_MT_CALLS.out.versions)
-        ch_versions = ch_versions.mix(GATK4_SELECTVARIANTS.out.versions)
+            POSTPROCESS_MT_CALLS(
+                CALL_SNV_MT.out.vcf,
+                CALL_SNV_MT_SHIFT.out.vcf,
+                ch_genome_fasta,
+                ch_genome_dictionary,
+                ch_genome_fai,
+                ch_mtshift_backchain,
+                ch_case_info,
+                ch_foundin_header,
+                ch_genome_chrsizes
+            )
+            ch_mt_vcf   = POSTPROCESS_MT_CALLS.out.vcf
+            ch_mt_tabix = POSTPROCESS_MT_CALLS.out.tbi
+            ch_versions = ch_versions.mix(CALL_SNV_MT.out.versions)
+            ch_versions = ch_versions.mix(CALL_SNV_MT_SHIFT.out.versions)
+            ch_versions = ch_versions.mix(POSTPROCESS_MT_CALLS.out.versions)
+            ch_versions = ch_versions.mix(GATK4_SELECTVARIANTS.out.versions)
+        }
 
     emit:
-        genome_vcf       = ch_genome_vcf                // channel: [ val(meta), path(vcf) ]
-        genome_tabix     = ch_genome_tabix              // channel: [ val(meta), path(tbi) ]
-        genome_vcf_tabix = ch_genome_vcf_tabix          // channel: [ val(meta), path(vcf), path(tbi) ]
-        genome_gvcf      = ch_gvcf                      // channel: [ val(meta), path(gvcf) ]
-        genome_gtabix    = ch_gtabix                    // channel: [ val(meta), path(gtbi) ]
-        mt_vcf           = POSTPROCESS_MT_CALLS.out.vcf // channel: [ val(meta), path(vcf) ]
-        mt_tabix         = POSTPROCESS_MT_CALLS.out.tbi // channel: [ val(meta), path(tbi) ]
-        versions         = ch_versions                  // channel: [ path(versions.yml) ]
+        genome_vcf       = ch_genome_vcf       // channel: [ val(meta), path(vcf) ]
+        genome_tabix     = ch_genome_tabix     // channel: [ val(meta), path(tbi) ]
+        genome_vcf_tabix = ch_genome_vcf_tabix // channel: [ val(meta), path(vcf), path(tbi) ]
+        genome_gvcf      = ch_gvcf             // channel: [ val(meta), path(gvcf) ]
+        genome_gtabix    = ch_gtabix           // channel: [ val(meta), path(gtbi) ]
+        mt_vcf           = ch_mt_vcf           // channel: [ val(meta), path(vcf) ]
+        mt_tabix         = ch_mt_tabix         // channel: [ val(meta), path(tbi) ]
+        versions         = ch_versions         // channel: [ path(versions.yml) ]
 }