merging lanes from baminput

nf-core · Jan 31, 2025 · 9b1cd04 · 9b1cd04
1 parent 0ceda2c
commit 9b1cd04
Show file tree

Hide file tree

Showing 2 changed files with 72 additions and 7 deletions.
diff --git a/subworkflows/local/merge_lanes_inputbam.nf b/subworkflows/local/merge_lanes_inputbam.nf
@@ -0,0 +1,58 @@
+//
+// Prepare reference indexing for downstream
+//
+
+include { SAMTOOLS_MERGE as SAMTOOLS_MERGE_LANES        } from '../../modules/nf-core/samtools/merge/main'
+include { SAMTOOLS_SORT  as SAMTOOLS_SORT_MERGED_LANES  } from '../../modules/nf-core/samtools/sort/main'
+include { SAMTOOLS_INDEX as SAMTOOLS_INDEX_MERGED_LANES } from '../../modules/nf-core/samtools/index/main'
+include { SAMTOOLS_FLAGSTAT as SAMTOOLS_FLAGSTAT_MAPPED } from '../../modules/nf-core/samtools/flagstat/main'
+
+workflow MERGE_LANES_INPUTBAM {
+    take:
+    bams // [ [meta], [bams] [bais]]
+
+    main:
+    ch_versions       = Channel.empty()
+    ch_multiqc_files  = Channel.empty()
+
+
+    ch_input_bam = bams
+
+    ch_input_for_lane_merge = bams
+                                .map { meta, bam, bai -> [ meta.clone().findAll{ it.key !in ['lane', 'colour_chemistry', 'shard_number'] }, bam ] }
+                                .groupTuple()
+                                .branch {
+                                    meta, bam ->
+                                        merge: bam.size() > 1
+                                        skip: true
+                                }
+
+    SAMTOOLS_MERGE_LANES ( ch_input_for_lane_merge.merge, [[], []], [[], []] )
+    ch_versions.mix( SAMTOOLS_MERGE_LANES.out.versions )
+
+    // Then mix back merged and single lane libraries for everything downstream
+    ch_mergedlanes_for_sorting = ch_input_for_lane_merge.skip
+                                    .mix( SAMTOOLS_MERGE_LANES.out.bam )
+
+    SAMTOOLS_SORT_MERGED_LANES ( ch_mergedlanes_for_sorting )
+    ch_mapped_bam = SAMTOOLS_SORT_MERGED_LANES.out.bam
+    ch_versions.mix( SAMTOOLS_SORT_MERGED_LANES.out.versions )
+
+    SAMTOOLS_INDEX_MERGED_LANES( ch_mapped_bam )
+    ch_mapped_bai =  params.fasta_largeref ? SAMTOOLS_INDEX_MERGED_LANES.out.csi : SAMTOOLS_INDEX_MERGED_LANES.out.bai
+    ch_versions.mix( SAMTOOLS_INDEX_MERGED_LANES.out.versions )
+
+    ch_input_for_flagstat = SAMTOOLS_SORT_MERGED_LANES.out.bam.join( SAMTOOLS_INDEX_MERGED_LANES.out.bai, failOnMismatch: true )
+
+    SAMTOOLS_FLAGSTAT_MAPPED ( ch_input_for_flagstat )
+    ch_versions.mix( SAMTOOLS_FLAGSTAT_MAPPED.out.versions.first() )
+    ch_multiqc_files = ch_multiqc_files.mix( SAMTOOLS_FLAGSTAT_MAPPED.out.flagstat )
+
+    emit:
+    bam        = ch_mapped_bam                            // [ [ meta ], bam ]
+    bai        = ch_mapped_bai                            // [ [ meta ], bai ]
+    flagstat   = SAMTOOLS_FLAGSTAT_MAPPED.out.flagstat    // [ [ meta ], stats ]
+    mqc        = ch_multiqc_files
+    versions   = ch_versions
+
+}
diff --git a/workflows/eager.nf b/workflows/eager.nf
@@ -24,6 +24,7 @@ include { addNewMetaFromAttributes } from '../subworkflows/local/utils_nfcore_ea
 include { REFERENCE_INDEXING            } from '../subworkflows/local/reference_indexing'
 include { PREPROCESSING                 } from '../subworkflows/local/preprocessing'
 include { MAP                           } from '../subworkflows/local/map'
+include { MERGE_LANES_INPUTBAM          } from '../subworkflows/local/merge_lanes_inputbam'
 include { FILTER_BAM                    } from '../subworkflows/local/bamfiltering.nf'
 include { DEDUPLICATE                   } from '../subworkflows/local/deduplicate'
 include { MANIPULATE_DAMAGE             } from '../subworkflows/local/manipulate_damage'
@@ -206,17 +207,23 @@ workflow EAGER {
 
         //
         // MODULE: flagstats of user supplied input BAMs
-        //
-        SAMTOOLS_FLAGSTATS_BAM_INPUT ( ch_bams_from_input )
-        ch_versions           = ch_versions.mix( SAMTOOLS_FLAGSTATS_BAM_INPUT.out.versions )
-        ch_flagstat_input_bam = SAMTOOLS_FLAGSTATS_BAM_INPUT.out.flagstat // For endorspy
+        // Maybe remove if subworkflow is already running it?
+        // SAMTOOLS_FLAGSTATS_BAM_INPUT ( ch_bams_from_input )
+        // ch_versions           = ch_versions.mix( SAMTOOLS_FLAGSTATS_BAM_INPUT.out.versions )
+        // ch_flagstat_input_bam = SAMTOOLS_FLAGSTATS_BAM_INPUT.out.flagstat // For endorspy
+
+        // SUBWORKFLOW: Merging lanes for ch_bams_from_input
 
+        MERGE_LANES_INPUTBAM(ch_bams_from_input)
+        ch_bams_from_input_lanemerged = MERGE_LANES_INPUTBAM.out.bam
+                                            .join(MERGE_LANES_INPUTBAM.out.bai)
 
     } else {
         ch_bams_from_input    = Channel.empty()
         ch_flagstat_input_bam = Channel.empty()
     }
 
+
     //
     // SUBWORKFLOW: bam filtering (length, mapped/unmapped, quality etc.)
     //
@@ -225,7 +232,7 @@ workflow EAGER {
 
         ch_mapped_for_bamfilter = MAP.out.bam
                                     .join(MAP.out.bai)
-                                    .mix(ch_bams_from_input)
+                                    .mix(ch_bams_from_input_lanemerged)
 
         FILTER_BAM ( ch_mapped_for_bamfilter )
         ch_bamfiltered_for_deduplication = FILTER_BAM.out.genomics
@@ -236,7 +243,7 @@ workflow EAGER {
     } else {
         ch_bamfiltered_for_deduplication = MAP.out.bam
                                                 .join(MAP.out.bai)
-                                                .mix(ch_bams_from_input)
+                                                .mix(ch_bams_from_input_lanemerged)
     }
 
     ch_reads_for_deduplication = ch_bamfiltered_for_deduplication
@@ -397,7 +404,7 @@ workflow EAGER {
     //
 
     ch_flagstat_for_endorspy_raw    = MAP.out.flagstat
-                                            .mix( ch_flagstat_input_bam )
+                                            .mix( MERGE_LANES_INPUTBAM.out.flagstat )
 
     if ( params.run_bamfiltering & !params.skip_deduplication ) {
         ch_for_endorspy = ch_flagstat_for_endorspy_raw