Update gzrt to v0.9.1 (#7241)

* Add module gzrt (#6918) * fixed environment.yml * Reply to review * gzrecover and gzip oneliner * gzrt uses a named pipe to avoid temp files * Updated test file URL * Improved syntax of prefixes in gzrt * Updated gzrt to 0.9.1 version * corrected name clash control in stub as well * updated biocontainers and singularity links * updated gzrt module * Fixed gzrt through named pipe * gzrt update snapshot * updated tests for gzrt * updated environment.yml * gzrt - updated snapshots * gzrt alpha - pre-release * Add args in command line - gzrt * gzrt - update snapshots and test dataset * should have not change it --------- Co-authored-by: tm4zza <[email protected]> Co-authored-by: Usman Rashid <[email protected]>
nf-core · Jan 7, 2025 · 56f433b · 56f433b
1 parent 7053802
commit 56f433b
Show file tree

Hide file tree

Showing 5 changed files with 211 additions and 46 deletions.
diff --git a/modules/nf-core/gzrt/environment.yml b/modules/nf-core/gzrt/environment.yml
@@ -3,4 +3,4 @@ channels:
   - conda-forge
   - bioconda
 dependencies:
-  - "bioconda::gzrt=0.8"
+  - "bioconda::gzrt=0.9.1"
diff --git a/modules/nf-core/gzrt/main.nf b/modules/nf-core/gzrt/main.nf
@@ -4,28 +4,52 @@ process GZRT {
 
     conda "${moduleDir}/environment.yml"
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/gzrt:0.8--he4a0461_0':
-        'biocontainers/gzrt:0.8--he4a0461_0' }"
+        'https://depot.galaxyproject.org/singularity/gzrt:0.9.1--h577a1d6_1':
+        'biocontainers/gzrt:0.9.1--h577a1d6_1' }"
 
     input:
     tuple val(meta), path(fastqgz)
 
     output:
-    tuple val(meta), path("*_recovered.fastq.gz"), emit: fastqrecovered
-    path "versions.yml"                          , emit: versions
+    tuple val(meta), path("${prefix}*.fastq.gz"), emit: recovered
+    path "versions.yml"                         , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
 
     script:
-    if (fastqgz.extension != "gz") {
-        error "GZRT works with .gz files only."
+    def args = task.ext.args ?: ''
+    prefix   = task.ext.prefix ?: "${meta.id}_recovered"
+    fastqgz.each { file ->
+        if (file.extension != "gz") {
+            error "GZRT works with .gz files only. Offending file: ${file}"
+        }
+
+        if ((meta.single_end && "${file}" == "${prefix}.fastq.gz") ||
+            (!meta.single_end && ("${file}" == "${prefix}_1.fastq.gz" || "${file}" == "${prefix}_2.fastq.gz"))) {
+                error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!"
+        }
     }
 
-    def prefix = task.ext.prefix ?: "${meta.id}"
     """
-    gzrecover -o ${prefix}_recovered.fastq ${fastqgz}
-    gzip ${prefix}_recovered.fastq
+    if [ "$meta.single_end" == true ]; then
+        gzrecover ${args} -p ${fastqgz} | gzip > ${prefix}.fastq.gz
+
+        if [ -e "${prefix}.fastq.gz" ] && [ ! -s "${prefix}.fastq.gz" ]; then
+            echo "" | gzip > ${prefix}.fastq.gz
+        fi
+    else
+        gzrecover ${args} -p ${fastqgz[0]} | gzip > ${prefix}_1.fastq.gz
+        gzrecover ${args} -p ${fastqgz[1]} | gzip > ${prefix}_2.fastq.gz
+
+        if [ -e "${prefix}_1.fastq.gz" ] && [ ! -s "${prefix}_1.fastq.gz" ]; then
+            echo "" | gzip > ${prefix}_1.fastq.gz
+        fi
+
+        if [ -e "${prefix}_2.fastq.gz" ] && [ ! -s "${prefix}_2.fastq.gz" ]; then
+            echo "" | gzip > ${prefix}_2.fastq.gz
+        fi
+    fi
 
     soft_line="${task.process}"
     ver_line="gzrt: \$(gzrecover -V |& sed '1!d ; s/gzrecover //')"
@@ -36,13 +60,27 @@ process GZRT {
     """
 
     stub:
-    def prefix = task.ext.prefix ?: "${meta.id}"
+    prefix = task.ext.prefix ?: "${meta.id}_recovered"
+    fastqgz.each { file ->
+        if (file.extension != "gz") {
+            error "GZRT works with .gz files only. Offending file: ${file}"
+        }
+
+        if ((meta.single_end && "${file}" == "${prefix}.fastq.gz") ||
+            (!meta.single_end && ("${file}" == "${prefix}_1.fastq.gz" || "${file}" == "${prefix}_2.fastq.gz"))) {
+                error "Input and output names are the same, use \"task.ext.prefix\" to disambiguate!"
+        }
+    }
     """
-    echo "" | gzip > ${prefix}_recovered.fastq.gz
+    if [ "$meta.single_end" == true ]; then
+        echo "" | gzip > ${prefix}.fastq.gz
+    else
+        echo "" | gzip > ${prefix}_1.fastq.gz
+        echo "" | gzip > ${prefix}_2.fastq.gz
+    fi
 
     soft_line="${task.process}"
     ver_line="gzrt: \$(gzrecover -V |& sed '1!d ; s/gzrecover //')"
-
     cat <<-END_VERSIONS > versions.yml
     "\${soft_line}":
         \${ver_line}

diff --git a/modules/nf-core/gzrt/meta.yml b/modules/nf-core/gzrt/meta.yml
@@ -23,21 +23,21 @@ input:
 
     - fastqgz:
         type: file
-        description: FASTQ.gz file
+        description: FASTQ.gz files
         pattern: "*.{gz}"
         ontologies:
           - edam: "http://edamontology.org/format_3989"
 
 output:
-  - fastqrecovered:
+  - recovered:
       - meta:
           type: map
           description: |
             Groovy Map containing sample information
             e.g. `[ id:'sample1', single_end:false ]`
-      - "*_recovered.fastq.gz":
+      - "${prefix}*.fastq.gz":
           type: file
-          description: Recovered FASTQ.gz file
+          description: Recovered FASTQ.gz files
           pattern: "*.{gz}"
           ontologies:
             - edam: "http://edamontology.org/format_3989"

diff --git a/modules/nf-core/gzrt/tests/main.nf.test b/modules/nf-core/gzrt/tests/main.nf.test
@@ -8,42 +8,87 @@ nextflow_process {
     tag "modules_nfcore"
     tag "gzrt"
 
-    test("Run gzrt on 30 paired-end reads - fastq.gz") {
-
+    test("Run gzrt on test2_1_corrupted single-end - fastq.gz") {
         when {
             process {
                 """
-                input[0] = [
-                    [ id:'test', single_end:false ], // meta map
-                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test_corrupted_30reads_R1.fastq.gz', checkIfExists: true)
+                input[0] =
+                [
+                    [ id:'test-single', single_end:true ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_1_corrupted_10kb.fastq.gz', checkIfExists: true)
                 ]
                 """
             }
         }
-
         then {
             assertAll(
                 { assert process.success },
                 { assert snapshot(process.out).match() }
             )
         }
+    }
 
+    test("Run gzrt on test2_1_corrupted and test2_2 paired-end - fastq.gz") {
+        when {
+            process {
+                """
+                input[0] =
+                [
+                    [ id:'test-paired', single_end:false ], // meta map
+                    [
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_1_corrupted_10kb.fastq.gz', checkIfExists: true),
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_2.fastq.gz', checkIfExists: true)
+                    ]
+                ]
+                """
+            }
+        }
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
     }
 
-    test("Run gzrt on 30 paired-end reads - fastq.gz - stub") {
+    test("Run gzrt on test2_1_corrupted single-end - fastq.gz - stub") {
         options "-stub"
 
         when {
             process {
                 """
                 input[0] = [
-                    [ id:'test', single_end:false ], // meta map
-                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test_corrupted_30reads_R1.fastq.gz', checkIfExists: true)
+                    [ id:'test-single-stub', single_end:true ], // meta map
+                    file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_1_corrupted_10kb.fastq.gz', checkIfExists: true)
                 ]
                 """
             }
         }
+        then {
+            assertAll(
+                { assert process.success },
+                { assert snapshot(process.out).match() }
+            )
+        }
+    }
+
+    test("Run gzrt on test2_1_corrupted and test2_2 paired-end - fastq.gz - stub") {
+        options "-stub"
 
+        when {
+            process {
+                """
+                input[0] =
+                [
+                    [ id:'test-paired-stub', single_end:false ], // meta map
+                    [
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_1_corrupted_10kb.fastq.gz', checkIfExists: true),
+                        file(params.modules_testdata_base_path + 'genomics/homo_sapiens/illumina/fastq/test2_2.fastq.gz', checkIfExists: true)
+                    ]
+                ]
+                """
+            }
+        }
         then {
             assertAll(
                 { assert process.success },

diff --git a/modules/nf-core/gzrt/tests/main.nf.test.snap b/modules/nf-core/gzrt/tests/main.nf.test.snap
@@ -1,72 +1,154 @@
 {
-    "Run gzrt on 30 paired-end reads - fastq.gz": {
+    "Run gzrt on test2_1_corrupted single-end - fastq.gz": {
         "content": [
             {
                 "0": [
                     [
                         {
-                            "id": "test",
+                            "id": "test-single",
+                            "single_end": true
+                        },
+                        "test-single_recovered.fastq.gz:md5,6bc7d0b6304039eadba7321d21e7b2c9"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
+                ],
+                "recovered": [
+                    [
+                        {
+                            "id": "test-single",
+                            "single_end": true
+                        },
+                        "test-single_recovered.fastq.gz:md5,6bc7d0b6304039eadba7321d21e7b2c9"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.3"
+        },
+        "timestamp": "2024-12-24T11:23:22.53088"
+    },
+    "Run gzrt on test2_1_corrupted single-end - fastq.gz - stub": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test-single-stub",
+                            "single_end": true
+                        },
+                        "test-single-stub_recovered.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
+                ],
+                "recovered": [
+                    [
+                        {
+                            "id": "test-single-stub",
+                            "single_end": true
+                        },
+                        "test-single-stub_recovered.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
+                ]
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.2",
+            "nextflow": "24.10.3"
+        },
+        "timestamp": "2024-12-24T11:24:13.88887"
+    },
+    "Run gzrt on test2_1_corrupted and test2_2 paired-end - fastq.gz - stub": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test-paired-stub",
                             "single_end": false
                         },
-                        "test_recovered.fastq.gz:md5,5eca10f954656478b5af96868222eec4"
+                        [
+                            "test-paired-stub_recovered_1.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940",
+                            "test-paired-stub_recovered_2.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                        ]
                     ]
                 ],
                 "1": [
-                    "versions.yml:md5,b9467d4cb860eb2941a2078a3dd3cf41"
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
                 ],
-                "fastqrecovered": [
+                "recovered": [
                     [
                         {
-                            "id": "test",
+                            "id": "test-paired-stub",
                             "single_end": false
                         },
-                        "test_recovered.fastq.gz:md5,5eca10f954656478b5af96868222eec4"
+                        [
+                            "test-paired-stub_recovered_1.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940",
+                            "test-paired-stub_recovered_2.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                        ]
                     ]
                 ],
                 "versions": [
-                    "versions.yml:md5,b9467d4cb860eb2941a2078a3dd3cf41"
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.2",
-            "nextflow": "24.10.1"
+            "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-15T13:49:10.174131"
+        "timestamp": "2024-12-24T11:24:58.632197"
     },
-    "Run gzrt on 30 paired-end reads - fastq.gz - stub": {
+    "Run gzrt on test2_1_corrupted and test2_2 paired-end - fastq.gz": {
         "content": [
             {
                 "0": [
                     [
                         {
-                            "id": "test",
+                            "id": "test-paired",
                             "single_end": false
                         },
-                        "test_recovered.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                        [
+                            "test-paired_recovered_1.fastq.gz:md5,6bc7d0b6304039eadba7321d21e7b2c9",
+                            "test-paired_recovered_2.fastq.gz:md5,641919946a0e572ca7483f6e54476f3b"
+                        ]
                     ]
                 ],
                 "1": [
-                    "versions.yml:md5,b9467d4cb860eb2941a2078a3dd3cf41"
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
                 ],
-                "fastqrecovered": [
+                "recovered": [
                     [
                         {
-                            "id": "test",
+                            "id": "test-paired",
                             "single_end": false
                         },
-                        "test_recovered.fastq.gz:md5,68b329da9893e34099c7d8ad5cb9c940"
+                        [
+                            "test-paired_recovered_1.fastq.gz:md5,6bc7d0b6304039eadba7321d21e7b2c9",
+                            "test-paired_recovered_2.fastq.gz:md5,641919946a0e572ca7483f6e54476f3b"
+                        ]
                     ]
                 ],
                 "versions": [
-                    "versions.yml:md5,b9467d4cb860eb2941a2078a3dd3cf41"
+                    "versions.yml:md5,6d1e28b8a8043e3cba67606c7acdc676"
                 ]
             }
         ],
         "meta": {
             "nf-test": "0.9.2",
-            "nextflow": "24.10.1"
+            "nextflow": "24.10.3"
         },
-        "timestamp": "2024-12-15T13:49:20.410351"
+        "timestamp": "2024-12-24T11:23:40.906933"
     }
 }