Fix join mismatch error for joint tumor only Mutect2

CHANGELOG.md

@@ -47,6 +47,7 @@ Rapaselet is a delta formed by the Rapaätno river between the Bielloriehppe mas
 - [#1173](https://github.com/nf-core/sarek/pull/1173) - Fixed duplicated entries in joint germline recalibrated VCF ([#966](https://github.com/nf-core/sarek/pull/966), [#1102](https://github.com/nf-core/sarek/pull/1102)),
   fixed grouping joint germline recalibrated VCF ([#1137](https://github.com/nf-core/sarek/pull/1137))
 - [#1177](https://github.com/nf-core/sarek/pull/1177) - Fix status inference when using nf-validation plugin
+- [#1181](https://github.com/nf-core/sarek/pull/1181) - Fix join mismatch error in Mutect2 tumor only subworkflow
 - [#1183](https://github.com/nf-core/sarek/pull/1183) - Add docs for concatentated germline variants
 - [#1184](https://github.com/nf-core/sarek/pull/1184) - Fix issue with duplicated variants in VCF from Sentieon-based joint-germline variant-calling with VQSR. (Corresponding to [#966](https://github.com/nf-core/sarek/issues/966) for GATK.)
 - [#1192](https://github.com/nf-core/sarek/pull/1192) - Add `ASCATprofile.png` to ASCAT output docs

subworkflows/local/bam_variant_calling_somatic_all/main.nf

@@ -70,8 +70,8 @@ workflow BAM_VARIANT_CALLING_SOMATIC_ALL {
     // CONTROLFREEC
     if (tools.split(',').contains('controlfreec')) {
         // Remap channels to match module/subworkflow
-        cram_normal = cram.map { meta, normal_cram, normal_crai, tumor_cram, tumor_crai -> [ meta, normal_cram, normal_crai ] }
-        cram_tumor = cram.map { meta, normal_cram, normal_crai, tumor_cram, tumor_crai -> [ meta, tumor_cram, tumor_crai ] }
+        cram_normal = cram.map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai -> [ meta, normal_cram, normal_crai ] }
+        cram_tumor = cram.map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai -> [ meta, tumor_cram, tumor_crai ] }
 
         MPILEUP_NORMAL(
             cram_normal,
@@ -183,7 +183,13 @@ workflow BAM_VARIANT_CALLING_SOMATIC_ALL {
     if (tools.split(',').contains('mutect2')) {
         BAM_VARIANT_CALLING_SOMATIC_MUTECT2(
             // Remap channel to match module/subworkflow
-            cram.map { meta, normal_cram, normal_crai, tumor_cram, tumor_crai -> [ meta, [ normal_cram, tumor_cram ], [ normal_crai, tumor_crai ] ] },
+            // Adjust meta.map to simplify joining channels
+            // joint_mutect2 mode needs different meta.map than regular mode
+            cram.map{ meta, normal_cram, normal_crai, tumor_cram, tumor_crai ->
+                joint_mutect2 ?
+                [ meta + [ id:meta.patient ] - meta.subMap('patient', 'tumor_id'), [ normal_cram, tumor_cram ], [ normal_crai, tumor_crai ] ] :
+                [ meta, [ normal_cram, tumor_cram ], [ normal_crai, tumor_crai ] ]
+            },
             // Remap channel to match module/subworkflow
             fasta.map{ it -> [ [ id:'fasta' ], it ] },
             // Remap channel to match module/subworkflow

subworkflows/local/bam_variant_calling_somatic_mutect2/main.nf

@@ -41,9 +41,9 @@ workflow BAM_VARIANT_CALLING_SOMATIC_MUTECT2 {
 
     if (joint_mutect2) {
         // Separate normal cram files and remove duplicates
-        ch_normal_cram = input.map{ meta, cram, crai -> [ meta - meta.subMap('tumor_id') + [id:meta.patient], cram[0], crai[0] ] }.unique()
+        ch_normal_cram = input.map{ meta, cram, crai -> [ meta, cram[0], crai[0] ] }.unique()
         // Extract tumor cram files
-        ch_tumor_cram = input.map{ meta, cram, crai -> [ meta - meta.subMap('tumor_id') + [id:meta.patient], cram[1], crai[1] ] }
+        ch_tumor_cram = input.map{ meta, cram, crai -> [ meta, cram[1], crai[1] ] }
         // Merge normal and tumor crams by patient
         ch_tn_cram = ch_normal_cram.mix(ch_tumor_cram).groupTuple()
         // Combine input and intervals for scatter and gather strategy
@@ -153,27 +153,27 @@ workflow BAM_VARIANT_CALLING_SOMATIC_MUTECT2 {
     CALCULATECONTAMINATION(ch_calculatecontamination_in_tables)
 
     // Initialize empty channel: Contamination calculation is run on pileup table, pileup is not run if germline resource is not provided
-    ch_seg_to_filtermutectcalls = Channel.empty()
-    ch_cont_to_filtermutectcalls = Channel.empty()
+    calculatecontamination_out_seg = Channel.empty()
+    calculatecontamination_out_cont = Channel.empty()
 
     if (joint_mutect2) {
         // Reduce the meta to only patient name
-        ch_seg_to_filtermutectcalls = CALCULATECONTAMINATION.out.segmentation.map{ meta, seg -> [ meta - meta.subMap('tumor_id') + [id: meta.patient], seg]}.groupTuple()
-        ch_cont_to_filtermutectcalls = CALCULATECONTAMINATION.out.contamination.map{ meta, cont -> [ meta - meta.subMap('tumor_id') + [id: meta.patient], cont]}.groupTuple()
+        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation.map{ meta, seg -> [ meta - meta.subMap('tumor_id') + [id: meta.patient], seg]}.groupTuple()
+        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination.map{ meta, cont -> [ meta - meta.subMap('tumor_id') + [id: meta.patient], cont]}.groupTuple()
     }
     else {
         // Keep tumor_vs_normal ID
-        ch_seg_to_filtermutectcalls = CALCULATECONTAMINATION.out.segmentation
-        ch_cont_to_filtermutectcalls = CALCULATECONTAMINATION.out.contamination
+        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation
+        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination
     }
 
     // Mutect2 calls filtered by filtermutectcalls using the artifactpriors, contamination and segmentation tables
     vcf_to_filter = vcf.join(tbi, failOnDuplicate: true, failOnMismatch: true)
-                        .join(stats, failOnDuplicate: true, failOnMismatch: true)
-                        .join(LEARNREADORIENTATIONMODEL.out.artifactprior, failOnDuplicate: true, failOnMismatch: true)
-                        .join(ch_seg_to_filtermutectcalls)
-                        .join(ch_cont_to_filtermutectcalls)
-                    .map{ meta, vcf, tbi, stats, orientation, seg, cont -> [ meta, vcf, tbi, stats, orientation, seg, cont, [] ] }
+        .join(stats, failOnDuplicate: true, failOnMismatch: true)
+        .join(LEARNREADORIENTATIONMODEL.out.artifactprior, failOnDuplicate: true, failOnMismatch: true)
+        .join(calculatecontamination_out_seg)
+        .join(calculatecontamination_out_cont)
+        .map{ meta, vcf, tbi, stats, orientation, seg, cont -> [ meta, vcf, tbi, stats, orientation, seg, cont, [] ] }
 
     FILTERMUTECTCALLS(vcf_to_filter, fasta, fai, dict)
 
@@ -205,8 +205,8 @@ workflow BAM_VARIANT_CALLING_SOMATIC_MUTECT2 {
     pileup_table_normal // channel: [ meta, table_normal ]
     pileup_table_tumor  // channel: [ meta, table_tumor ]
 
-    contamination_table    = ch_cont_to_filtermutectcalls    // channel: [ meta, contamination ]
-    segmentation_table     = ch_seg_to_filtermutectcalls     // channel: [ meta, segmentation ]
+    contamination_table    = calculatecontamination_out_cont    // channel: [ meta, contamination ]
+    segmentation_table     = calculatecontamination_out_seg     // channel: [ meta, segmentation ]
 
     versions // channel: [ versions.yml ]
 }

subworkflows/local/bam_variant_calling_tumor_only_all/main.nf

@@ -110,7 +110,13 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_ALL {
     // MUTECT2
     if (tools.split(',').contains('mutect2')) {
         BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2(
-            cram,
+            // Adjust meta.map to simplify joining channels
+            // joint_mutect2 mode needs different meta.map than regular mode
+            cram.map{ meta, cram, crai ->
+                joint_mutect2 ?
+                [ meta + [ id:meta.patient ] - meta.subMap('sample', 'status', 'num_intervals', 'data_type', 'patient') , cram, crai ] :
+                [ meta - meta.subMap('sample', 'status', 'num_intervals', 'data_type'), cram, crai ]
+            },
             // Remap channel to match module/subworkflow
             fasta.map{ it -> [ [ id:'fasta' ], it ] },
             // Remap channel to match module/subworkflow

subworkflows/local/bam_variant_calling_tumor_only_mutect2/main.nf

@@ -41,7 +41,7 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2 {
     if (joint_mutect2) {
         // Perform variant calling using mutect2 module in tumor single mode
         // Group cram files by patient
-        patient_crams = input.map{ meta, t_cram, t_crai -> [ meta - meta.subMap('sample') + [id:meta.patient], t_cram, t_crai ] }.groupTuple()
+        patient_crams = input.groupTuple()
         // Add intervals for scatter-gather scaling
         patient_cram_intervals = patient_crams.combine(intervals)
         // Move num_intervals to meta map and reorganize channel for MUTECT2 module
@@ -90,10 +90,11 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2 {
     MERGEMUTECTSTATS(stats_to_merge)
 
     // Mix intervals and no_intervals channels together
-    vcf   = Channel.empty().mix(MERGE_MUTECT2.out.vcf, vcf_branch.no_intervals)
-    tbi   = Channel.empty().mix(MERGE_MUTECT2.out.tbi, tbi_branch.no_intervals)
-    stats = Channel.empty().mix(MERGEMUTECTSTATS.out.stats, stats_branch.no_intervals)
-    f1r2  = Channel.empty().mix(f1r2_to_merge, f1r2_branch.no_intervals)
+    // Remove unnecessary metadata
+    vcf   = Channel.empty().mix(MERGE_MUTECT2.out.vcf, vcf_branch.no_intervals).map{ meta, vcf -> [ meta - meta.subMap('num_intervals'), vcf ] }
+    tbi   = Channel.empty().mix(MERGE_MUTECT2.out.tbi, tbi_branch.no_intervals).map{ meta, tbi -> [ meta - meta.subMap('num_intervals'), tbi ] }
+    stats = Channel.empty().mix(MERGEMUTECTSTATS.out.stats, stats_branch.no_intervals).map{ meta, stats -> [ meta - meta.subMap('num_intervals'), stats ] }
+    f1r2  = Channel.empty().mix(f1r2_to_merge, f1r2_branch.no_intervals).map{ meta, f1r2 -> [ meta - meta.subMap('num_intervals'), f1r2 ] }
 
     // Generate artifactpriors using learnreadorientationmodel on the f1r2 output of mutect2
     LEARNREADORIENTATIONMODEL(f1r2)
@@ -124,14 +125,13 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2 {
     calculatecontamination_out_cont = Channel.empty()
 
     if (joint_mutect2) {
-        // Remove sample names and retain patient name as the main identifier
-        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation.map{ meta, seg -> [ meta - meta.subMap('sample') + [id:meta.patient], seg ] }.groupTuple()
-        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination.map{ meta, cont -> [ meta - meta.subMap('sample') + [id:meta.patient], cont ] }.groupTuple()
-    }
-    else {
+        // Group tables by samples
+        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation.map{ meta, seg -> [ meta - meta.subMap('num_intervals'), seg ] }.groupTuple()
+        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination.map{ meta, cont -> [ meta - meta.subMap('num_intervals'), cont ] }.groupTuple()
+    } else {
         // Regular single sample mode
-        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation
-        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination
+        calculatecontamination_out_seg = CALCULATECONTAMINATION.out.segmentation.map{ meta, seg -> [ meta - meta.subMap('num_intervals'), seg ] }
+        calculatecontamination_out_cont = CALCULATECONTAMINATION.out.contamination.map{ meta, cont -> [ meta - meta.subMap('num_intervals'), cont ] }
     }
 
     // Mutect2 calls filtered by filtermutectcalls using the contamination and segmentation tables
@@ -146,7 +146,7 @@ workflow BAM_VARIANT_CALLING_TUMOR_ONLY_MUTECT2 {
 
     vcf_filtered = FILTERMUTECTCALLS.out.vcf
         // add variantcaller to meta map and remove no longer necessary field: num_intervals
-        .map{ meta, vcf -> [ meta - meta.subMap('num_intervals') + [ variantcaller:'mutect2' ], vcf ] }
+        .map{ meta, vcf -> [ meta + [ variantcaller:'mutect2' ], vcf ] }
 
     versions = versions.mix(MERGE_MUTECT2.out.versions)
     versions = versions.mix(CALCULATECONTAMINATION.out.versions)