Merge pull request #1135 from nf-core/avoid_generating_redundant_index

Avoid redundant index generation in sentieon flow
nf-core · Jun 27, 2023 · df0f578 · df0f578
2 parents f387cf3 + 78979b0
commit df0f578
Show file tree

Hide file tree

Showing 3 changed files with 14 additions and 7 deletions.
diff --git a/subworkflows/local/bam_sentieon_dedup/main.nf b/subworkflows/local/bam_sentieon_dedup/main.nf
@@ -5,14 +5,12 @@
 // A when clause condition is defined in the conf/modules.config to determine if the module should be run
 
 include { CRAM_QC_MOSDEPTH_SAMTOOLS              } from '../cram_qc_mosdepth_samtools/main'
-include { GATK4_MARKDUPLICATES                   } from '../../../modules/nf-core/gatk4/markduplicates/main'
 include { SENTIEON_DEDUP                         } from '../../../modules/nf-core/sentieon/dedup/main'
-include { SAMTOOLS_INDEX as INDEX_INPUT          } from '../../../modules/nf-core/samtools/index/main'
-include { SAMTOOLS_INDEX as INDEX_MARKDUPLICATES } from '../../../modules/nf-core/samtools/index/main'
 
 workflow BAM_SENTIEON_DEDUP {
     take:
     bam                    // channel: [mandatory] [ meta, bam ]  // Although the channel is named "bam", it may contain cram-files.
+    bai
     fasta                  // channel: [mandatory] [ fasta ]
     fasta_fai              // channel: [mandatory] [ fasta_fai ]
     intervals_bed_combined // channel: [optional]  [ intervals_bed ]
@@ -21,9 +19,9 @@ workflow BAM_SENTIEON_DEDUP {
     versions = Channel.empty()
     reports  = Channel.empty()
 
-    INDEX_INPUT(bam)
-    bam_bai = bam.join(INDEX_INPUT.out.bai.concat(INDEX_INPUT.out.crai), failOnMismatch:true, failOnDuplicate:true)
-    // The concat operation is part of the above command since if the "bam" channel contains cram-files, then the index files will be in the channel INDEX_INPUT.out.crai and not in INDEX_INPUT.out.bai
+    bam = bam.map{ meta, bam -> [ meta - meta.subMap('data_type'), bam ] }
+    bai = bai.map{ meta, bai -> [ meta - meta.subMap('data_type'), bai ] }
+    bam_bai = bam.join(bai, failOnMismatch:true, failOnDuplicate:true)
     SENTIEON_DEDUP(bam_bai, fasta, fasta_fai)
 
     // Join with the crai file

diff --git a/subworkflows/local/fastq_align_bwamem_mem2_dragmap_sentieon/main.nf b/subworkflows/local/fastq_align_bwamem_mem2_dragmap_sentieon/main.nf
@@ -36,6 +36,9 @@ workflow FASTQ_ALIGN_BWAMEM_MEM2_DRAGMAP_SENTIEON {
     bam = bam.mix(DRAGMAP_ALIGN.out.bam)
     bam = bam.mix(SENTIEON_BWAMEM.out.bam_and_bai.map{ meta, bam, bai -> [ meta, bam ] })
 
+    bai = Channel.empty()
+    bai = bai.mix(SENTIEON_BWAMEM.out.bam_and_bai.map{ meta, bam, bai -> [ meta, bai ] })
+
     // Gather reports of all tools used
     reports = reports.mix(DRAGMAP_ALIGN.out.log)
 
@@ -47,6 +50,7 @@ workflow FASTQ_ALIGN_BWAMEM_MEM2_DRAGMAP_SENTIEON {
 
     emit:
     bam      // channel: [ [meta], bam ]
+    bai      // channel: [ [meta], bai ]
     reports
     versions // channel: [ versions.yml ]
 }
diff --git a/workflows/sarek.nf b/workflows/sarek.nf
@@ -540,6 +540,10 @@ workflow SAREK {
             [ groupKey( meta - meta.subMap('num_lanes', 'read_group', 'size') + [ data_type:'bam', id:meta.sample ], (meta.num_lanes ?: 1) * (meta.size ?: 1)), bam ]
         }.groupTuple()
 
+        bai_mapped = FASTQ_ALIGN_BWAMEM_MEM2_DRAGMAP_SENTIEON.out.bai.map{ meta, bai ->
+            [ groupKey( meta - meta.subMap('num_lanes', 'read_group', 'size') + [ data_type:'bai', id:meta.sample ], (meta.num_lanes ?: 1) * (meta.size ?: 1)), bai ]
+        }.groupTuple()
+
         // gatk4 markduplicates can handle multiple bams as input, so no need to merge/index here
         // Except if and only if save_mapped or (skipping markduplicates and sentieon-dedup)
         if (
@@ -578,7 +582,6 @@ workflow SAREK {
         // ch_bam_for_markduplicates will contain bam mapped with FASTQ_ALIGN_BWAMEM_MEM2_DRAGMAP_SENTIEON when step is mapping
         // Or bams that are specified in the samplesheet.csv when step is prepare_recalibration
         cram_for_markduplicates = params.step == 'mapping' ? bam_mapped : input_sample.map{ meta, input, index -> [ meta, input ] }
-
         // if no MD is done, then run QC on mapped & converted CRAM files
         // or the input BAM (+converted) or CRAM files
         cram_skip_markduplicates = Channel.empty()
@@ -628,8 +631,10 @@ workflow SAREK {
             // Gather used softwares versions
             versions = versions.mix(BAM_MARKDUPLICATES_SPARK.out.versions)
         } else if (params.tools && params.tools.split(',').contains('sentieon_dedup')) {
+            crai_for_markduplicates = params.step == 'mapping' ? bai_mapped : input_sample.map{ meta, input, index -> [ meta, index ] }
             BAM_SENTIEON_DEDUP(
                 cram_for_markduplicates,
+                crai_for_markduplicates,
                 fasta,
                 fasta_fai,
                 intervals_for_preprocessing)