Merge pull request #142 from sanger-tol/dev

DLBPointon · web-flow · commit dcd6e07e3218 · 2026-03-10T13:26:42.000Z
Dev2Main
diff --git a/modules.json b/modules.json
@@ -99,7 +99,7 @@
                     },
                     "samtools/merge": {
                         "branch": "master",
-                        "git_sha": "b2e78932ef01165fd85829513eaca29eff8e640a",
+                        "git_sha": "440edf75d8782913115a7b72a88392a227f72cc1",
                         "installed_by": ["bam_samtools_merge_markdup", "modules"]
                     },
                     "samtools/sort": {
@@ -184,7 +184,7 @@
                     },
                     "samtools/mergedup": {
                         "branch": "main",
-                        "git_sha": "2e28234fb0e2970767dab28310c7eb553ad40081",
+                        "git_sha": "55f3dd54f274c9648207a82bf68d257945d27053",
                         "installed_by": ["bam_samtools_merge_markdup"]
                     },
                     "telomere/extract": {
@@ -208,12 +208,12 @@
                 "sanger-tol": {
                     "bam_samtools_merge_markdup": {
                         "branch": "main",
-                        "git_sha": "2e28234fb0e2970767dab28310c7eb553ad40081",
+                        "git_sha": "55f3dd54f274c9648207a82bf68d257945d27053",
                         "installed_by": ["cram_map_illumina_hic"]
                     },
                     "cram_map_illumina_hic": {
                         "branch": "main",
-                        "git_sha": "2e28234fb0e2970767dab28310c7eb553ad40081",
+                        "git_sha": "55f3dd54f274c9648207a82bf68d257945d27053",
                         "installed_by": ["subworkflows"]
                     },
                     "gap_finder": {
diff --git a/modules/nf-core/samtools/merge/main.nf b/modules/nf-core/samtools/merge/main.nf
diff --git a/modules/nf-core/samtools/merge/meta.yml b/modules/nf-core/samtools/merge/meta.yml
diff --git a/modules/nf-core/samtools/merge/tests/main.nf.test b/modules/nf-core/samtools/merge/tests/main.nf.test
diff --git a/modules/sanger-tol/samtools/mergedup/main.nf b/modules/sanger-tol/samtools/mergedup/main.nf
@@ -8,8 +8,8 @@ process SAMTOOLS_MERGEDUP {
         'community.wave.seqera.io/library/htslib_samtools:1.23--cde2c40a51d6f752' }"
 
     input:
-    tuple val(meta) , path(input)
-    tuple val(meta2), path(fasta), path(fai),  path(gzi)
+    tuple val(meta) , path(input_files, stageAs: "?/*"), path(index_files, stageAs: "?/*")
+    tuple val(meta2), path(fasta), path(fai), path(gzi)
 
     output:
     tuple val(meta), path("*.bam")      , emit: bam,  optional: true
@@ -32,9 +32,11 @@ process SAMTOOLS_MERGEDUP {
                     "bam"
     """
     samtools merge \\
+        --threads ${task.cpus-1} \\
         ${args} \\
         - \\
-        ${input} |\\
+        ${reference} \\
+        ${input_files} |\\
     samtools markdup \\
         -T ${prefix} \\
         -f ${prefix}.metrics \\
diff --git a/modules/sanger-tol/samtools/mergedup/meta.yml b/modules/sanger-tol/samtools/mergedup/meta.yml
@@ -40,11 +40,16 @@ input:
         description: |
           Groovy Map containing sample information
           e.g. [ id:'test', single_end:false ]
-    - input:
+    - input_files:
         type: file
         description: BAM/CRAM/SAM files
         pattern: "*.{bam,cram,sam}"
         ontologies: []
+    - index_files:
+        type: file
+        description: BAM/CRAM/SAM index
+        pattern: "*.{bai,csi,crai}"
+        ontologies: []
   - - meta2:
         type: map
         description: |
diff --git a/modules/sanger-tol/samtools/mergedup/tests/main.nf.test b/modules/sanger-tol/samtools/mergedup/tests/main.nf.test
@@ -32,7 +32,8 @@ nextflow_process {
                     [
                         file(params.modules_testdata_base_path + 'Undibacterium_unclassified/analysis/hic_mapping/41741_2.7.sub.cram.baUndUnlc1_hic_phased_hap1.2.bam', checkIfExists: true),
                         file(params.modules_testdata_base_path + 'Undibacterium_unclassified/analysis/hic_mapping/41741_2.7.sub.cram.baUndUnlc1_hic_phased_hap1.3.bam', checkIfExists: true),
-                    ]
+                    ],
+                    []
                 ])
                 input[1] = Channel.of([[:],[],[],[]])
                 """
@@ -72,7 +73,8 @@ nextflow_process {
                     [
                         file(params.modules_testdata_base_path + 'Undibacterium_unclassified/analysis/hic_mapping/41741_2.7.sub.cram.baUndUnlc1_hic_phased_hap1.2.bam', checkIfExists: true),
                         file(params.modules_testdata_base_path + 'Undibacterium_unclassified/analysis/hic_mapping/41741_2.7.sub.cram.baUndUnlc1_hic_phased_hap1.3.bam', checkIfExists: true),
-                    ]
+                    ],
+                    []
                 ])
                 input[1] = Channel.of([[:],[],[],[]])
                 """
diff --git a/modules/sanger-tol/samtools/mergedup/tests/main.nf.test.snap b/modules/sanger-tol/samtools/mergedup/tests/main.nf.test.snap
@@ -68,11 +68,11 @@
                 ]
             }
         ],
+        "timestamp": "2026-02-09T15:16:40.761606",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-09T15:16:40.761606"
+        }
     },
     "sarscov2 - bam": {
         "content": [
@@ -96,10 +96,10 @@
                 ]
             }
         ],
+        "timestamp": "2026-02-09T15:16:36.500954",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-09T15:16:36.500954"
+        }
     }
 }
diff --git a/subworkflows/local/longread_coverage/main.nf b/subworkflows/local/longread_coverage/main.nf
@@ -51,7 +51,8 @@ workflow LONGREAD_COVERAGE {
         .map { bams ->
             tuple (
                 [ id    : bams.first().name.split('_').first() ], // Change sample ID
-                bams
+                bams,
+                []
             )
         }
         .set { collected_files_for_merge }
diff --git a/subworkflows/sanger-tol/bam_samtools_merge_markdup/main.nf b/subworkflows/sanger-tol/bam_samtools_merge_markdup/main.nf
@@ -36,7 +36,7 @@ workflow BAM_SAMTOOLS_MERGE_MARKDUP {
         .combine(ch_assemblies, by: 0)
         .combine(ch_fai_gzi, by: 0)
         .multiMap { meta, bams, assembly, fai, gzi ->
-            bam:   [ meta, bams ]
+            bam:   [ meta, bams, [] ]
             fasta: [ meta, assembly, fai, gzi ]
         }
 
diff --git a/subworkflows/sanger-tol/cram_map_illumina_hic/main.nf b/subworkflows/sanger-tol/cram_map_illumina_hic/main.nf
@@ -39,14 +39,13 @@ workflow CRAM_MAP_ILLUMINA_HIC {
         .map { meta, cram -> [ meta + [ cramfile: cram ], cram ]}
 
     ch_hic_cram_raw = ch_hic_cram_meta_mod
-        .branch { meta, cram ->
-            def cram_file = file(cram, checkIfExists: true)
-            def index = cram + ".crai"
-            have_index: file(index).exists()
-                return [ meta, cram_file, file(index, checkIfExists: true) ]
-            no_index: true
-                return [ meta, cram_file ]
-        }
+            .branch { meta, cram ->
+                def index = file(cram.toUriString() + ".crai")
+                have_index: index.exists()
+                    return [ meta, cram, index ]
+                no_index: true
+                    return [ meta, cram ]
+            }
 
     //
     // Module: Index CRAM files without indexes
diff --git a/subworkflows/sanger-tol/cram_map_illumina_hic/tests/main.nf.test.snap b/subworkflows/sanger-tol/cram_map_illumina_hic/tests/main.nf.test.snap
@@ -26,11 +26,11 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:43:38.748216",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:43:38.748216"
+        }
     },
     "meles meles - bwamem2": {
         "content": [
@@ -59,11 +59,11 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:39:34.946622",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:39:34.946622"
+        }
     },
     "meles meles - minimap2": {
         "content": [
@@ -92,11 +92,11 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:42:47.872589",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:42:47.872589"
+        }
     },
     "meles meles - zymo - bwamem2 - multiple samples - double files": {
         "content": [
@@ -143,11 +143,11 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:41:09.214803",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:41:09.214803"
+        }
     },
     "meles meles - multiple read groups - bwamem2": {
         "content": [
@@ -176,11 +176,11 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:45:13.292929",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:45:13.292929"
+        }
     },
     "meles meles - bwamem2 - multiple samples - meta mismatch": {
         "content": [
@@ -231,10 +231,10 @@
                 ]
             ]
         ],
+        "timestamp": "2026-02-10T10:42:06.647189",
         "meta": {
             "nf-test": "0.9.3",
             "nextflow": "25.10.3"
-        },
-        "timestamp": "2026-02-10T10:42:06.647189"
+        }
     }
 }

Original file line number	Diff line number	Diff line change
`@@ -68,11 +68,11 @@`
`68`	`68`	`]`
`69`	`69`	`}`
`70`	`70`	`],`
	`71`	`+ "timestamp": "2026-02-09T15:16:40.761606",`
`71`	`72`	`"meta": {`
`72`	`73`	`"nf-test": "0.9.3",`
`73`	`74`	`"nextflow": "25.10.3"`
`74`		`- },`
`75`		`- "timestamp": "2026-02-09T15:16:40.761606"`
	`75`	`+ }`
`76`	`76`	`},`
`77`	`77`	`"sarscov2 - bam": {`
`78`	`78`	`"content": [`
`@@ -96,10 +96,10 @@`
`96`	`96`	`]`
`97`	`97`	`}`
`98`	`98`	`],`
	`99`	`+ "timestamp": "2026-02-09T15:16:36.500954",`
`99`	`100`	`"meta": {`
`100`	`101`	`"nf-test": "0.9.3",`
`101`	`102`	`"nextflow": "25.10.3"`
`102`		`- },`
`103`		`- "timestamp": "2026-02-09T15:16:36.500954"`
	`103`	`+ }`
`104`	`104`	`}`
`105`	`105`	`}`
Original file line number	Diff line number	Diff line change
`@@ -51,7 +51,8 @@ workflow LONGREAD_COVERAGE {`
`51`	`51`	`.map { bams ->`
`52`	`52`	`tuple (`
`53`	`53`	`[ id : bams.first().name.split('_').first() ], // Change sample ID`
`54`		`- bams`
	`54`	`+ bams,`
	`55`	`+ []`
`55`	`56`	`)`
`56`	`57`	`}`
`57`	`58`	`.set { collected_files_for_merge }`
Original file line number	Diff line number	Diff line change
`@@ -36,7 +36,7 @@ workflow BAM_SAMTOOLS_MERGE_MARKDUP {`
`36`	`36`	`.combine(ch_assemblies, by: 0)`
`37`	`37`	`.combine(ch_fai_gzi, by: 0)`
`38`	`38`	`.multiMap { meta, bams, assembly, fai, gzi ->`
`39`		`- bam: [ meta, bams ]`
	`39`	`+ bam: [ meta, bams, [] ]`
`40`	`40`	`fasta: [ meta, assembly, fai, gzi ]`
`41`	`41`	`}`
`42`	`42`