MarkDuplicates Input specificity and input refactor.

modules/nf-core/gatk4/markduplicates/main.nf

@@ -13,25 +13,24 @@ process GATK4_MARKDUPLICATES {
     path  fasta_fai
 
     output:
-    tuple val(meta), path("*cram"),     emit: cram,  optional: true
-    tuple val(meta), path("*bam"),      emit: bam,   optional: true
-    tuple val(meta), path("*.crai"),    emit: crai,  optional: true
-    tuple val(meta), path("*.bai"),     emit: bai,   optional: true
-    tuple val(meta), path("*.metrics"), emit: metrics
-    path "versions.yml",                emit: versions
+    tuple val(meta), path("${prefix}.cram"),    emit: cram,  optional: true
+    tuple val(meta), path("${prefix}.bam"),     emit: bam,   optional: true
+    tuple val(meta), path("${prefix}*crai"),    emit: crai,  optional: true
+    tuple val(meta), path("${prefix}*bai"),     emit: bai,   optional: true
+    tuple val(meta), path("*.metrics"),         emit: metrics
+    path "versions.yml",                        emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
     def args = task.ext.args ?: ''
-    prefix = task.ext.prefix ?: "${meta.id}.bam"
-
-    // If the extension is CRAM, then change it to BAM
-    prefix_bam = prefix.tokenize('.')[-1] == 'cram' ? "${prefix.substring(0, prefix.lastIndexOf('.'))}.bam" : prefix
+    prefix = task.ext.prefix ?: "${meta.id}.md"
+    suffix = task.ext.suffix ?: "bam"
 
     def input_list = bam.collect{"--INPUT $it"}.join(' ')
-    def reference = fasta ? "--REFERENCE_SEQUENCE ${fasta}" : ""
+    file_extension = bam.first().getExtension()
+    reference = fasta ? "--REFERENCE_SEQUENCE ${fasta}" : ""
 
     def avail_mem = 3072
     if (!task.memory) {
@@ -46,17 +45,17 @@ process GATK4_MARKDUPLICATES {
     gatk --java-options "-Xmx${avail_mem}M -XX:-UsePerfData" \\
         MarkDuplicates \\
         $input_list \\
-        --OUTPUT ${prefix_bam} \\
+        --OUTPUT ${prefix}.bam \\
         --METRICS_FILE ${prefix}.metrics \\
         --TMP_DIR . \\
         ${reference} \\
         $args
-
+    
     # If cram files are wished as output, the run samtools for conversion
-    if [[ ${prefix} == *.cram ]]; then
-        samtools view -Ch -T ${fasta} -o ${prefix} ${prefix_bam}
-        rm ${prefix_bam}
-        samtools index ${prefix}
+    if [[ ${suffix}  == cram ]]; then
+        samtools view -Ch -T ${fasta} -o ${prefix}.cram ${prefix}.bam 
+        rm ${prefix}.bam
+        samtools index ${prefix}.cram
     fi
 
     cat <<-END_VERSIONS > versions.yml

modules/nf-core/gatk4/markduplicates/meta.yml

@@ -43,49 +43,64 @@ output:
           description: |
             Groovy Map containing sample information
             e.g. [ id:'test', single_end:false ]
-      - "*cram":
-          type: file
-          description: Marked duplicates CRAM file
+          pattern: "*.{cram}"
+      - ${prefix}.cram:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
           pattern: "*.{cram}"
   - bam:
       - meta:
           type: map
           description: |
             Groovy Map containing sample information
             e.g. [ id:'test', single_end:false ]
-      - "*bam":
-          type: file
-          description: Marked duplicates BAM file
+          pattern: "*.{bam}"
+      - ${prefix}.bam:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
           pattern: "*.{bam}"
   - crai:
       - meta:
           type: map
           description: |
             Groovy Map containing sample information
             e.g. [ id:'test', single_end:false ]
-      - "*.crai":
-          type: file
-          description: CRAM index file
+          pattern: "*.{cram.crai}"
+      - ${prefix}*crai:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
           pattern: "*.{cram.crai}"
   - bai:
       - meta:
           type: map
           description: |
             Groovy Map containing sample information
             e.g. [ id:'test', single_end:false ]
-      - "*.bai":
-          type: file
-          description: BAM index file
+          pattern: "*.{bam.bai}"
+      - ${prefix}*bai:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
           pattern: "*.{bam.bai}"
   - metrics:
       - meta:
           type: map
           description: |
             Groovy Map containing sample information
             e.g. [ id:'test', single_end:false ]
+          pattern: "*.{metrics.txt}"
       - "*.metrics":
-          type: file
-          description: Duplicate metrics file generated by GATK
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
           pattern: "*.{metrics.txt}"
   - versions:
       - versions.yml:

modules/nf-core/gatk4/markduplicates/tests/bam.config

@@ -2,7 +2,6 @@ process {
 
     withName: GATK4_MARKDUPLICATES {
         ext.args = '--CREATE_INDEX true'
-        ext.prefix = { "${meta.id}.bam" }
     }
 
 }

modules/nf-core/gatk4/markduplicates/tests/cram.config

@@ -2,7 +2,7 @@ process {
 
     withName: GATK4_MARKDUPLICATES {
         ext.args = '--CREATE_INDEX true'
-        ext.prefix = { "${meta.id}.cram" }
+        ext.suffix  = "cram"
     }
 
 }

modules/nf-core/gatk4/markduplicates/tests/main.nf.test

@@ -28,7 +28,7 @@ nextflow_process {
         then {
             assertAll(
                 { assert process.success },
-                { assert snapshot(process.out.bam).match("bam") },
+                { assert snapshot(bam(process.out.bam[0][1]).getReadsMD5())},
                 { assert snapshot(process.out.bai).match("bai") },
                 { assert snapshot(process.out.versions).match("versions") },
                 { assert snapshot(file(process.out.metrics[0][1]).name).match("test.metrics") }
@@ -58,8 +58,8 @@ nextflow_process {
         then {
             assertAll(
                 { assert process.success },
-                { assert snapshot(process.out.bam).match("multi bam") },
-                { assert snapshot(process.out.bai).match("multi bai") },
+                { assert snapshot(bam(process.out.bam[0][1]).getReadsMD5())},
+                { assert snapshot(file(process.out.bai[0][1]).name).match("multi bai") },
                 { assert snapshot(process.out.versions).match("multi versions") },
                 { assert snapshot(file(process.out.metrics[0][1]).name).match("multi test.metrics") }
             )

modules/nf-core/gatk4/markduplicates/tests/main.nf.test.snap

@@ -1,49 +1,23 @@
 {
-    "multi bam": {
-        "content": [
-            [
-                [
-                    {
-                        "id": "test",
-                        "single_end": false
-                    },
-                    "test.bam:md5,8a808b1a94d2627c4d659a2151c4cb9f"
-                ]
-            ]
-        ],
-        "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
-        },
-        "timestamp": "2024-02-13T15:21:36.059923"
-    },
     "multi crai": {
         "content": [
-            "test.cram.crai"
+            "test.md.cram.crai"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2023-12-12T17:43:37.780426007"
+        "timestamp": "2025-03-28T14:12:43.343356201"
     },
     "multi bai": {
         "content": [
-            [
-                [
-                    {
-                        "id": "test",
-                        "single_end": false
-                    },
-                    "test.bai:md5,38b99c5f771895ecf5324c3186b9d452"
-                ]
-            ]
+            "test.md.bai"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2024-02-13T15:21:36.09642"
+        "timestamp": "2025-03-28T14:12:29.424717865"
     },
     "versions": {
         "content": [
@@ -52,20 +26,20 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2024-02-13T15:21:08.710549"
+        "timestamp": "2025-03-28T14:12:15.803964407"
     },
     "multi test.metrics": {
         "content": [
-            "test.bam.metrics"
+            "test.md.metrics"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2023-12-12T17:43:11.732892667"
+        "timestamp": "2025-03-28T14:12:29.438583878"
     },
     "bai": {
         "content": [
@@ -75,15 +49,15 @@
                         "id": "test",
                         "single_end": false
                     },
-                    "test.bai:md5,26001bcdbce12e9f07557d8f7b8d360e"
+                    "test.md.bai:md5,7a2b177be9e77a3a8ef1baa1afb5aef4"
                 ]
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2023-12-12T17:42:39.651888758"
+        "timestamp": "2025-03-28T14:12:15.788812713"
     },
     "multi cram versions": {
         "content": [
@@ -92,10 +66,10 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2024-02-13T15:21:56.966376"
+        "timestamp": "2025-03-28T14:12:43.351899261"
     },
     "multi versions": {
         "content": [
@@ -104,57 +78,39 @@
             ]
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2024-02-13T15:21:36.138095"
+        "timestamp": "2025-03-28T14:12:29.432440299"
     },
     "multi cram test.metrics": {
         "content": [
-            "test.cram.metrics"
+            "test.md.metrics"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2023-12-12T17:43:37.798977444"
+        "timestamp": "2025-03-28T14:12:43.361537975"
     },
     "multi cram": {
         "content": [
-            "test.cram"
-        ],
-        "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
-        },
-        "timestamp": "2023-12-12T17:43:37.771137858"
-    },
-    "bam": {
-        "content": [
-            [
-                [
-                    {
-                        "id": "test",
-                        "single_end": false
-                    },
-                    "test.bam:md5,75d914ba8804eaf2acf02ab432197ec9"
-                ]
-            ]
+            "test.md.cram"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2024-02-13T15:21:08.645892"
+        "timestamp": "2025-03-28T14:12:43.332947793"
     },
     "test.metrics": {
         "content": [
-            "test.bam.metrics"
+            "test.md.metrics"
         ],
         "meta": {
-            "nf-test": "0.8.4",
-            "nextflow": "23.10.0"
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.5"
         },
-        "timestamp": "2023-12-12T17:42:39.672508385"
+        "timestamp": "2025-03-28T14:12:15.810908919"
     }
 }