Add dsh-bio export-segments module (#631)

Co-authored-by: Gregor Sturm <mail@gregor-sturm.de>

modules/dshbio/exportsegments/functions.nf

@@ -0,0 +1,68 @@
+//
+//  Utility functions used in nf-core DSL2 module files
+//
+
+//
+// Extract name of software tool from process name using $task.process
+//
+def getSoftwareName(task_process) {
+    return task_process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()
+}
+
+//
+// Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+//
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args            = args.args ?: ''
+    options.args2           = args.args2 ?: ''
+    options.args3           = args.args3 ?: ''
+    options.publish_by_meta = args.publish_by_meta ?: []
+    options.publish_dir     = args.publish_dir ?: ''
+    options.publish_files   = args.publish_files
+    options.suffix          = args.suffix ?: ''
+    return options
+}
+
+//
+// Tidy up and join elements of a list to return a path string
+//
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }      // Remove empty entries
+    paths     = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", "") } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+//
+// Function to save/publish module results
+//
+def saveFiles(Map args) {
+    if (!args.filename.endsWith('.version.txt')) {
+        def ioptions  = initOptions(args.options)
+        def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+        if (ioptions.publish_by_meta) {
+            def key_list = ioptions.publish_by_meta instanceof List ? ioptions.publish_by_meta : args.publish_by_meta
+            for (key in key_list) {
+                if (args.meta && key instanceof String) {
+                    def path = key
+                    if (args.meta.containsKey(key)) {
+                        path = args.meta[key] instanceof Boolean ? "${key}_${args.meta[key]}".toString() : args.meta[key]
+                    }
+                    path = path instanceof String ? path : ''
+                    path_list.add(path)
+                }
+            }
+        }
+        if (ioptions.publish_files instanceof Map) {
+            for (ext in ioptions.publish_files) {
+                if (args.filename.endsWith(ext.key)) {
+                    def ext_list = path_list.collect()
+                    ext_list.add(ext.value)
+                    return "${getPathFromList(ext_list)}/$args.filename"
+                }
+            }
+        } else if (ioptions.publish_files == null) {
+            return "${getPathFromList(path_list)}/$args.filename"
+        }
+    }
+}

modules/dshbio/exportsegments/main.nf

@@ -0,0 +1,40 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process DSHBIO_EXPORTSEGMENTS {
+    tag "${meta.id}"
+    label 'process_medium'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['id']) }
+
+    conda (params.enable_conda ? "bioconda::dsh-bio=2.0.5" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/dsh-bio:2.0.5--hdfd78af_0"
+    } else {
+        container "quay.io/biocontainers/dsh-bio:2.0.5--hdfd78af_0"
+    }
+
+    input:
+    tuple val(meta), path(gfa)
+
+    output:
+    tuple val(meta), path("*.fa"), emit: fasta
+    path "*.version.txt"             , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    dsh-bio \\
+        export-segments \\
+        $options.args \\
+        -i $gfa \\
+        -o ${prefix}.fa
+
+    echo \$(dsh-bio --version 2>&1) | grep -o 'dsh-bio-tools .*' | cut -f2 -d ' ' > ${software}.version.txt
+    """
+}

modules/dshbio/exportsegments/meta.yml

@@ -0,0 +1,40 @@
+name: dshbio_exportsegments
+description: Export assembly segment sequences in GFA 1.0 format to FASTA format
+keywords:
+    - gfa
+    - assembly
+    - segment
+tools:
+    - dshbio:
+        description: |
+            Reads, features, variants, assemblies, alignments, genomic range trees, pangenome
+            graphs, and a bunch of random command line tools for bioinformatics. LGPL version 3
+            or later.
+        homepage: https://github.com/heuermh/dishevelled-bio
+        documentation: https://github.com/heuermh/dishevelled-bio
+input:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - gfa:
+        type: file
+        description: Assembly segments in GFA 1.0 format
+        pattern: "*.{gfa}"
+output:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - fasta:
+        type: file
+        description: Assembly segment sequences in FASTA format
+        pattern: "*.{fa}"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@heuermh"

tests/config/pytest_modules.yml

@@ -250,6 +250,10 @@ dragonflye:
   - modules/dragonflye/**
   - tests/modules/dragonflye/**
 
+dshbio/exportsegments:
+  - modules/dshbio/exportsegments/**
+  - tests/modules/dshbio/exportsegments/**
+
 dshbio/filterbed:
   - modules/dshbio/filterbed/**
   - tests/modules/dshbio/filterbed/**

tests/modules/dshbio/exportsegments/main.nf

@@ -0,0 +1,13 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { DSHBIO_EXPORTSEGMENTS } from '../../../../modules/dshbio/exportsegments/main.nf' addParams( options: [:] )
+
+workflow test_dshbio_exportsegments {
+    input = [ [ id:'test' ], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['assembly_gfa'], checkIfExists: true) ]
+            ]
+
+    DSHBIO_EXPORTSEGMENTS ( input )
+}

tests/modules/dshbio/exportsegments/test.yml

@@ -0,0 +1,8 @@
+- name: dshbio exportsegments
+  command: nextflow run ./tests/modules/dshbio/exportsegments -entry test_dshbio_exportsegments -c tests/config/nextflow.config
+  tags:
+    - dshbio
+    - dshbio/exportsegments
+  files:
+    - path: ./output/dshbio/test.fa
+      md5sum: 19ed0b69970ed3fbb641c5c510ebef61