Update WIPERTOOLS (#7300)

* Refactor fastq and report variable names for consistency across wipertools modules * Refactor variable names for consistency in fastq and report processing modules * Update test assertions to reflect changes in output variable names in fastqwiper module * Update test assertions to match changes in report output structure in fastqwiper tests
nf-core · Jan 14, 2025 · c9e9241 · c9e9241
1 parent fb1f03c
commit c9e9241
Show file tree

Hide file tree

Showing 15 changed files with 129 additions and 71 deletions.
diff --git a/modules/nf-core/wipertools/fastqgather/main.nf b/modules/nf-core/wipertools/fastqgather/main.nf
@@ -8,28 +8,28 @@ process WIPERTOOLS_FASTQGATHER {
         'biocontainers/wipertools:1.1.3--pyhdfd78af_0' }"
 
     input:
-    tuple val(meta), path(fastq_in)
+    tuple val(meta), path(fastq)
 
     output:
-    tuple val(meta), path("${prefix}.fastq.gz"), emit: fastq_out
+    tuple val(meta), path("${prefix}.fastq.gz"), emit: gathered_fastq
     path "versions.yml"                        , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
-    def args    = task.ext.args ?: ''
-    prefix      = task.ext.prefix ?: "${meta.id}_gather"
+    def args = task.ext.args ?: ''
+    prefix   = task.ext.prefix ?: "${meta.id}_gather"
 
     // Check if the output file name is in the list of input files
-    if (fastq_in.any { it.name == "${prefix}.fastq.gz" }) {
+    if (fastq.any { it.name == "${prefix}.fastq.gz" }) {
         error 'Output file name "${prefix}.fastq.gz}" matches one of the input files. Use \"task.ext.prefix\" to disambiguate!.'
     }
 
     """
     wipertools \\
         fastqgather \\
-        -i $fastq_in \\
+        -i $fastq \\
         -o ${prefix}.fastq.gz \\
         ${args}
 
@@ -43,7 +43,7 @@ process WIPERTOOLS_FASTQGATHER {
     prefix      = task.ext.prefix ?: "${meta.id}_gather"
 
     // Check if the output file name is in the list of input files
-    if (fastq_in.any { it.name == "${prefix}.fastq.gz" }) {
+    if (fastq.any { it.name == "${prefix}.fastq.gz" }) {
         error 'Output file name "${prefix}.fastq.gz}" matches one of the input files. Use \"task.ext.prefix\" to disambiguate!.'
     }
     """

diff --git a/modules/nf-core/wipertools/fastqgather/meta.yml b/modules/nf-core/wipertools/fastqgather/meta.yml
@@ -7,8 +7,7 @@ keywords:
   - union
 tools:
   - "fastqgather":
-      description:
-        "A tool of the wipertools suite that merges FASTQ chunks produced\
+      description: "A tool of the wipertools suite that merges FASTQ chunks produced\
         \ by wipertools_fastqscatter."
       homepage: "https://github.com/mazzalab/fastqwiper"
       documentation: "https://github.com/mazzalab/fastqwiper"
@@ -24,15 +23,15 @@ input:
         description: |
           Groovy Map containing sample information
           e.g. [ id:'test' ]
-    - fastq_in:
+    - fastq:
         type: file
         description: List of FASTQ chunk files to be merged
         pattern: "*.{fastq,fastq.gz}"
         ontologies:
           - edam: "http://edamontology.org/format_1930"
           - edam: "http://edamontology.org/format_3989"
 output:
-  - fastq_out:
+  - gathered_fastq:
       - meta:
           type: map
           description: |

diff --git a/modules/nf-core/wipertools/fastqgather/tests/main.nf.test b/modules/nf-core/wipertools/fastqgather/tests/main.nf.test
@@ -33,7 +33,7 @@ nextflow_process {
             }
             process {
                 """
-                input[0] = WIPERTOOLS_FASTQSCATTER.out.chunks
+                input[0] = WIPERTOOLS_FASTQSCATTER.out.fastq_chunks
                 """
             }
         }
@@ -56,7 +56,7 @@ nextflow_process {
             }
             process {
                 """
-                input[0] = WIPERTOOLS_FASTQSCATTER.out.chunks
+                input[0] = WIPERTOOLS_FASTQSCATTER.out.fastq_chunks
                 """
             }
         }

diff --git a/modules/nf-core/wipertools/fastqgather/tests/main.nf.test.snap b/modules/nf-core/wipertools/fastqgather/tests/main.nf.test.snap
@@ -13,7 +13,7 @@
                 "1": [
                     "versions.yml:md5,291a2e6d58146bfcf57b4d9ef850cbd9"
                 ],
-                "fastq_out": [
+                "gathered_fastq": [
                     [
                         {
                             "id": "fastq_chunks"
@@ -30,7 +30,7 @@
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:28:22.882325"
+        "timestamp": "2025-01-13T18:49:15.441781"
     },
     "merge two gzipped fastq - fastq.gz": {
         "content": [
@@ -46,7 +46,7 @@
                 "1": [
                     "versions.yml:md5,291a2e6d58146bfcf57b4d9ef850cbd9"
                 ],
-                "fastq_out": [
+                "gathered_fastq": [
                     [
                         {
                             "id": "fastq_chunks"
@@ -63,6 +63,6 @@
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:28:12.396668"
+        "timestamp": "2025-01-13T18:49:02.891972"
     }
 }
diff --git a/modules/nf-core/wipertools/fastqscatter/main.nf b/modules/nf-core/wipertools/fastqscatter/main.nf
@@ -12,7 +12,7 @@ process WIPERTOOLS_FASTQSCATTER {
     val(num_splits)
 
     output:
-    tuple val(meta), path("${out_folder}/*") , emit: chunks
+    tuple val(meta), path("${out_folder}/*") , emit: fastq_chunks
     path "versions.yml"                      , emit: versions
 
     when:
@@ -24,6 +24,9 @@ process WIPERTOOLS_FASTQSCATTER {
     def args_list   = args.tokenize()
     out_folder      = (args_list.contains('--out_folder') ? args_list[args_list.indexOf('--out_folder')+1] :
                         (args_list.contains('-o') ? args_list[args_list.indexOf('-o')+1] : 'chunks'))
+    if(!args.contains('-o') && !args.contains('--out_folder')) {
+        args += " -o ${out_folder}"
+    }
     """
     wipertools \\
         fastqscatter \\
@@ -44,6 +47,9 @@ process WIPERTOOLS_FASTQSCATTER {
     def args_list = args.tokenize()
     out_folder    = (args_list.contains('--out_folder') ? args_list[args_list.indexOf('--out_folder')+1] :
                         (args_list.contains('-o') ? args_list[args_list.indexOf('-o')+1] : 'chunks'))
+    if(!args.contains('-o') && !args.contains('--out_folder')) {
+        args += " -o ${out_folder}"
+    }
     """
     mkdir ${out_folder}
     for i in {1..${num_splits}}

diff --git a/modules/nf-core/wipertools/fastqscatter/meta.yml b/modules/nf-core/wipertools/fastqscatter/meta.yml
@@ -34,7 +34,7 @@ input:
         pattern: "[1-9][0-9]*"
 
 output:
-  - chunks:
+  - fastq_chunks:
       - meta:
           type: map
           description: |

diff --git a/modules/nf-core/wipertools/fastqscatter/tests/main.nf.test.snap b/modules/nf-core/wipertools/fastqscatter/tests/main.nf.test.snap
@@ -16,7 +16,7 @@
                 "1": [
                     "versions.yml:md5,c8e701c1048cc9ab79752d1bf631ad4b"
                 ],
-                "chunks": [
+                "fastq_chunks": [
                     [
                         {
                             "id": "test"
@@ -36,7 +36,7 @@
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:28:32.330894"
+        "timestamp": "2025-01-13T18:41:58.317846"
     },
     "3 reads, one truncated - fastq - stub": {
         "content": [
@@ -55,7 +55,7 @@
                 "1": [
                     "versions.yml:md5,c8e701c1048cc9ab79752d1bf631ad4b"
                 ],
-                "chunks": [
+                "fastq_chunks": [
                     [
                         {
                             "id": "test"
@@ -75,6 +75,6 @@
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:28:41.304878"
+        "timestamp": "2025-01-13T18:42:09.268891"
     }
 }
diff --git a/modules/nf-core/wipertools/fastqwiper/main.nf b/modules/nf-core/wipertools/fastqwiper/main.nf
@@ -8,11 +8,11 @@ process WIPERTOOLS_FASTQWIPER {
         'biocontainers/wipertools:1.1.3--pyhdfd78af_0' }"
 
     input:
-    tuple val(meta), path(fastq_in)
+    tuple val(meta), path(fastq)
 
     output:
-    tuple val(meta), path("${prefix}.fastq.gz"), emit: fastq_out
-    path("*.report")                           , emit: report
+    tuple val(meta), path("${prefix}.fastq.gz"), emit: wiped_fastq
+    tuple val(meta), path("*.report")          , emit: report
     path "versions.yml"                        , emit: versions
 
     when:
@@ -21,11 +21,11 @@ process WIPERTOOLS_FASTQWIPER {
     script:
     def args    = task.ext.args ?: ''
     prefix = task.ext.prefix ?: "${meta.id}_wiped"
-    if ("${fastq_in}" == "${prefix}.fastq.gz") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!."
+    if ("${fastq}" == "${prefix}.fastq.gz") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!."
     """
     wipertools \\
         fastqwiper \\
-        -i ${fastq_in} \\
+        -i ${fastq} \\
         -o ${prefix}.fastq.gz \\
         -r ${prefix}.report \\
         ${args}
@@ -38,7 +38,7 @@ process WIPERTOOLS_FASTQWIPER {
 
     stub:
     prefix = task.ext.prefix ?: "${meta.id}_wiped"
-    if ("${fastq_in}" == "${prefix}.fastq.gz") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!."
+    if ("${fastq}" == "${prefix}.fastq.gz") error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!."
     """
     echo "" | gzip > ${prefix}.fastq.gz
     touch ${prefix}.report

diff --git a/modules/nf-core/wipertools/fastqwiper/meta.yml b/modules/nf-core/wipertools/fastqwiper/meta.yml
@@ -23,15 +23,15 @@ input:
         description: |
           Groovy Map containing sample information
           e.g. [ id:'test' ]
-    - fastq_in:
+    - fastq:
         type: file
         description: FASTQ file to be cleaned
         pattern: "*.{fastq,fastq.gz}"
         ontologies:
           - edam: "http://edamontology.org/format_1930"
           - edam: "http://edamontology.org/format_3989"
 output:
-  - fastq_out:
+  - wiped_fastq:
       - meta:
           type: map
           description: |
@@ -45,6 +45,11 @@ output:
             - edam: "http://edamontology.org/format_1930"
             - edam: "http://edamontology.org/format_3989"
   - report:
+      - meta:
+          type: map
+          description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test' ]
       - "*.report":
           type: file
           description: Summary of the cleaning task

diff --git a/modules/nf-core/wipertools/fastqwiper/tests/main.nf.test b/modules/nf-core/wipertools/fastqwiper/tests/main.nf.test
@@ -28,18 +28,18 @@ nextflow_process {
         then {
             assertAll(
                 { assert process.success },
-                { assert path(process.out.fastq_out.get(0).get(1)).linesGzip.size() == 8},
+                { assert path(process.out.wiped_fastq.get(0).get(1)).linesGzip.size() == 8},
                 {
-                    def read_final_qual = path(process.out.fastq_out.get(0).get(1)).linesGzip[7]
+                    def read_final_qual = path(process.out.wiped_fastq.get(0).get(1)).linesGzip[7]
                     assert read_final_qual.equals('CCCCCGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGGG')
                 },
-                { assert path(process.out.report.get(0)).readLines().size() == 11},
+                { assert path(process.out.report.get(0).get(1)).readLines().size() == 11},
                 {
-                    def report_header = path(process.out.report.get(0)).readLines()[0]
+                    def report_header = path(process.out.report.get(0).get(1)).readLines()[0]
                     assert report_header.equals('FASTQWIPER REPORT:')
                 },
                 {
-                    def qual_check = path(process.out.report.get(0)).readLines()[9]
+                    def qual_check = path(process.out.report.get(0).get(1)).readLines()[9]
                     assert qual_check.equals('Reads discarded because len(SEQ) neq len(QUAL): 1')
                 },
                 { assert snapshot(process.out).match() }

diff --git a/modules/nf-core/wipertools/fastqwiper/tests/main.nf.test.snap b/modules/nf-core/wipertools/fastqwiper/tests/main.nf.test.snap
@@ -11,32 +11,42 @@
                     ]
                 ],
                 "1": [
-                    "test_wiped.report:md5,d41d8cd98f00b204e9800998ecf8427e"
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test_wiped.report:md5,d41d8cd98f00b204e9800998ecf8427e"
+                    ]
                 ],
                 "2": [
                     "versions.yml:md5,0a814cc53368590314bd01d1b433b6fd"
                 ],
-                "fastq_out": [
+                "report": [
                     [
                         {
                             "id": "test"
                         },
-                        "test_wiped.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                        "test_wiped.report:md5,d41d8cd98f00b204e9800998ecf8427e"
                     ]
                 ],
-                "report": [
-                    "test_wiped.report:md5,d41d8cd98f00b204e9800998ecf8427e"
-                ],
                 "versions": [
                     "versions.yml:md5,0a814cc53368590314bd01d1b433b6fd"
+                ],
+                "wiped_fastq": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test_wiped.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                    ]
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:29:03.969824"
+        "timestamp": "2025-01-13T18:52:39.908974"
     },
     "Quality mismatch - fastq": {
         "content": [
@@ -50,31 +60,41 @@
                     ]
                 ],
                 "1": [
-                    "test_wiped.report:md5,d6fe1f0066f538993bb8f6b90150c8d5"
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test_wiped.report:md5,d6fe1f0066f538993bb8f6b90150c8d5"
+                    ]
                 ],
                 "2": [
                     "versions.yml:md5,0a814cc53368590314bd01d1b433b6fd"
                 ],
-                "fastq_out": [
+                "report": [
                     [
                         {
                             "id": "test"
                         },
-                        "test_wiped.fastq.gz:md5,81b1d99cf5c1615123a8c31760858a10"
+                        "test_wiped.report:md5,d6fe1f0066f538993bb8f6b90150c8d5"
                     ]
                 ],
-                "report": [
-                    "test_wiped.report:md5,d6fe1f0066f538993bb8f6b90150c8d5"
-                ],
                 "versions": [
                     "versions.yml:md5,0a814cc53368590314bd01d1b433b6fd"
+                ],
+                "wiped_fastq": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test_wiped.fastq.gz:md5,81b1d99cf5c1615123a8c31760858a10"
+                    ]
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.2",
             "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-17T20:28:53.747935"
+        "timestamp": "2025-01-13T18:52:27.962146"
     }
 }