add nf-core style module for assembly_manifest

CHANGELOG.md

@@ -9,6 +9,9 @@ Initial release of nf-core/seqsubmit, created with the [nf-core](https://nf-co.r
 
 ### `Added`
 
+- [#12](https://github.com/nf-core/seqsubmit/pull/12)
+  - Add initial version of `GENERATE_ASSEMBLY_MANIFEST` (by @fxquah)
+
 - [#2](https://github.com/nf-core/proteinfamilies/pull/2)
   - Added initial versions of local modules for `GENOME_UPLOAD` and `ENA_WEBIN_CLI`. (by @KateSakharova and @vagkaratzas)
   - Added initial version of the `GENOMESUBMIT` workflow. (by @KateSakharova and @vagkaratzas)

assets/samplesheet.csv

@@ -1,3 +1,3 @@
 sample,fasta,accession,assembly_software,binning_software,binning_parameters,stats_generation_software,completeness,contamination,genome_coverage,metagenome,co-assembly,broad_environment,local_environment,environmental_medium,rRNA_presence,NCBI_lineage
 lachnospira_eligens,https://github.com/nf-core/test-datasets/raw/seqsubmit/test_data/bins/bin_lachnospira_eligens.fa.gz,SRR24458089,spades_v3.15.5,mags_v1,default,CheckM2_v1.0.1,61.0,0.21,32.07,sediment metagenome,False,marine,cable bacteria,marine sediment,False,d__Bacteria;p__Proteobacteria;c__Deltaproteobacteria;o__Desulfobacterales;f__Desulfobulbaceae;g__Candidatus Electrothrix;s__
-lachnospiraceae,https://github.com/nf-core/test-datasets/raw/seqsubmit/test_data/bins/bin_lachnospiraceae.fa.gz,SRR24458087,spades_v3.15.5,mags_v1,default,CheckM2_v1.0.1,92.81,1.09,66.04,sediment metagenome,False,marine,cable bacteria,marine sediment,False,d__Bacteria;p__Proteobacteria;c__Deltaproteobacteria;o__Desulfobacterales;f__Desulfobulbaceae;g__Candidatus Electrothrix;s__Candidatus Electrothrix marina
+lachnospiraceae,https://github.com/nf-core/test-datasets/raw/seqsubmit/test_data/bins/bin_lachnospiraceae.fa.gz,SRR24458087,spades_v3.15.5,mags_v1,default,CheckM2_v1.0.1,92.81,1.09,66.04,sediment metagenome,False,marine,cable bacteria,marine sediment,False,d__Bacteria;p__Proteobacteria;c__Deltaproteobacteria;o__Desulfobacterales;f__Desulfobulbaceae;g__Candidatus Electrothrix;s__Candidatus Electrothrix marina

modules/local/generate_assembly_manifest/environment.yml

@@ -0,0 +1,6 @@
+---
+channels:
+  - conda-forge
+  - bioconda
+dependencies:
+  - "bioconda::assembly_uploader=1.3.2"

modules/local/generate_assembly_manifest/main.nf

@@ -0,0 +1,47 @@
+process GENERATE_ASSEMBLY_MANIFEST {
+    tag "$meta.id"
+    label 'process_single'
+
+    container "community.wave.seqera.io/library/pip_assembly-uploader:7e9461afbdd7a521"
+
+    input:
+    tuple val(meta), path(assembly_fasta), path(data_csv)
+    val(assembly_study)
+
+    output:
+    tuple val(meta), path("${assembly_study}_upload/*.manifest") , emit: manifest
+    path "versions.yml", emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    assembly_manifest \\
+        --study ${assembly_study} \\
+        --data ${data_csv} \\
+        --assembly_study ${assembly_study} \\
+        --output ${assembly_study}_upload \\
+        ${args}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        python: \$(python --version)
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    mkdir ${assembly_study}_upload/
+    touch ${assembly_study}_upload/test.manifest
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        python: \$(python --version)
+    END_VERSIONS
+    """
+}

modules/local/generate_assembly_manifest/meta.yml

@@ -0,0 +1,50 @@
+name: "generate_assembly_manifest"
+description: Generate manifest files for genome assemblies using assembly-uploader
+keywords:
+  - ena
+  - manifest
+  - upload
+tools:
+  - assembly-uploader:
+      description: Make a manifest file for each assembly.
+        Co-assemblies cannot be generated from a mix of private and public runs -
+        all runs used in a co-assembly must have the same privacy status (all private or all public).
+      documentation: https://github.com/EBI-Metagenomics/assembly_uploader/
+      licence: ["Apache License"]
+input:
+  - - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information.
+          e.g. [ id:'test', single_end:false ]
+    - assembly_fasta:
+        type: file
+        description: Genome assembly FASTA file.
+    - data_csv:
+        type: file
+        description: |
+          CSV file containing metadata -
+          runs (comma-separated and in quotes, example: "SRR1234,SRR5678"),
+          coverage, assembler, version, filepath and optionally sample.
+    - assembly_study:
+        type: value
+        description: |
+          Pre-existing study ID to submit to if available.
+          Must exist in the webin account.
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information.
+        e.g. [ id:'test', single_end:false ]
+  - manifest:
+      type: file
+      description: Manifest file required for submission with webin-cli.
+  - versions:
+      type: file
+      description: File containing software versions.
+      pattern: "versions.yml"
+
+authors:
+  - "@fxquah"

modules/local/generate_assembly_manifest/tests/main.nf.test

@@ -0,0 +1,73 @@
+nextflow_process {
+
+    name "Test Process GENERATE_ASSEMBLY_MANIFEST"
+    script "../main.nf"
+    process "GENERATE_ASSEMBLY_MANIFEST"
+
+    tag "modules"
+    tag "modules_"
+    tag "generate_assembly_manifest"
+
+    test("GENERATE_ASSEMBLY_MANIFEST completes with expected outputs") {
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id:'test' ],
+                    file(params.pipelines_testdata_base_path + "/test_data/contigs.fasta.gz", checkIfExists: true),
+                    file(params.pipelines_testdata_base_path + "/samplesheets/samplesheet_generatemanifest.csv", checkIfExists: true)
+                ]
+                input[1] = "PRJ12345"
+                """
+
+            }
+        }
+
+        then {
+            assert process.success
+            assertAll(
+                { assert snapshot(
+                    process.out,
+                    path(process.out.versions[0]).yaml
+                ).match() },
+                { assert process.out.manifest.size() == 1 },
+                { assert process.out.manifest[0][1].toString().endsWith(".manifest") },
+                {
+                    def manifestContent = path(process.out.manifest[0][1]).text
+                    assert manifestContent.contains("PRJ12345") : "PRJ12345 not found in manifest file"
+                }
+            )
+        }
+
+    }
+
+    test("GENERATE_ASSEMBLY_MANIFEST completes with expected outputs - stub") {
+
+        options "-stub"
+
+        when {
+            process {
+                """
+                input[0] = [
+                    [ id:'test' ],
+                    file('test_contigs.fastq.gz'),
+                    file('test_samplesheet_generatemanifest.csv')
+                ]
+                input[1] = "PRJ12345"
+                """
+            }
+        }
+
+        then {
+            assert process.success
+            assertAll(
+                { assert snapshot(
+                    process.out,
+                    path(process.out.versions[0]).yaml
+                ).match() }
+            )
+        }
+
+    }
+
+}

modules/local/generate_assembly_manifest/tests/main.nf.test.snap

@@ -0,0 +1,78 @@
+{
+    "GENERATE_ASSEMBLY_MANIFEST completes with expected outputs - stub": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test.manifest:md5,d41d8cd98f00b204e9800998ecf8427e"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,188d7f56d019104ee5883d69eede27b4"
+                ],
+                "manifest": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "test.manifest:md5,d41d8cd98f00b204e9800998ecf8427e"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,188d7f56d019104ee5883d69eede27b4"
+                ]
+            },
+            {
+                "GENERATE_ASSEMBLY_MANIFEST": {
+                    "python": "Python 3.14.0"
+                }
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.3",
+            "nextflow": "25.10.0"
+        },
+        "timestamp": "2025-10-29T11:00:23.072784"
+    },
+    "GENERATE_ASSEMBLY_MANIFEST completes with expected outputs": {
+        "content": [
+            {
+                "0": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "233126d4c4d023f18c7836ed36395e3c.manifest:md5,b31fec00db575da44ad17950ae2e37ff"
+                    ]
+                ],
+                "1": [
+                    "versions.yml:md5,188d7f56d019104ee5883d69eede27b4"
+                ],
+                "manifest": [
+                    [
+                        {
+                            "id": "test"
+                        },
+                        "233126d4c4d023f18c7836ed36395e3c.manifest:md5,b31fec00db575da44ad17950ae2e37ff"
+                    ]
+                ],
+                "versions": [
+                    "versions.yml:md5,188d7f56d019104ee5883d69eede27b4"
+                ]
+            },
+            {
+                "GENERATE_ASSEMBLY_MANIFEST": {
+                    "python": "Python 3.14.0"
+                }
+            }
+        ],
+        "meta": {
+            "nf-test": "0.9.3",
+            "nextflow": "25.10.0"
+        },
+        "timestamp": "2025-10-29T11:00:16.801535"
+    }
+}