add new assembly scan module (#971)

* add new assembly scan module

* add newline endings

* fix newline ending

* add newline at end

Co-authored-by: Peri <rrx8@cdc.gov>
Co-authored-by: Robert A. Petit III <robbie.petit@gmail.com>
Co-authored-by: Gregor Sturm <mail@gregor-sturm.de>

modules/assemblyscan/functions.nf

@@ -0,0 +1,78 @@
+//
+//  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()
+}
+
+//
+// Extract name of module from process name using $task.process
+//
+def getProcessName(task_process) {
+    return task_process.tokenize(':')[-1]
+}
+
+//
+// Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+//
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args            = args.args ?: ''
+    options.args2           = args.args2 ?: ''
+    options.args3           = args.args3 ?: ''
+    options.publish_by_meta = args.publish_by_meta ?: []
+    options.publish_dir     = args.publish_dir ?: ''
+    options.publish_files   = args.publish_files
+    options.suffix          = args.suffix ?: ''
+    return options
+}
+
+//
+// Tidy up and join elements of a list to return a path string
+//
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }      // Remove empty entries
+    paths     = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", "") } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+//
+// Function to save/publish module results
+//
+def saveFiles(Map args) {
+    def ioptions  = initOptions(args.options)
+    def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+
+    // Do not publish versions.yml unless running from pytest workflow
+    if (args.filename.equals('versions.yml') && !System.getenv("NF_CORE_MODULES_TEST")) {
+        return null
+    }
+    if (ioptions.publish_by_meta) {
+        def key_list = ioptions.publish_by_meta instanceof List ? ioptions.publish_by_meta : args.publish_by_meta
+        for (key in key_list) {
+            if (args.meta && key instanceof String) {
+                def path = key
+                if (args.meta.containsKey(key)) {
+                    path = args.meta[key] instanceof Boolean ? "${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/assemblyscan/main.nf

@@ -0,0 +1,38 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process ASSEMBLYSCAN {
+    tag "$meta.id"
+    label 'process_low'
+    publishDir "${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::assembly-scan=0.4.1" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/assembly-scan:0.4.1--pyhdfd78af_0"
+    } else {
+        container "quay.io/biocontainers/assembly-scan:0.4.1--pyhdfd78af_0"
+    }
+
+    input:
+    tuple val(meta), path(assembly)
+
+    output:
+    tuple val(meta), path("*.json"), emit: json
+    path "versions.yml"            , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    assembly-scan $assembly > ${prefix}.json
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$( assembly-scan --version 2>&1 | sed 's/^.*assembly-scan //; s/Using.*\$//' )
+    END_VERSIONS
+    """
+}

modules/assemblyscan/meta.yml

@@ -0,0 +1,43 @@
+name: assemblyscan
+description: Assembly summary statistics in JSON format
+keywords:
+  - assembly
+  - statistics
+tools:
+  - assemblyscan:
+      description: Assembly summary statistics in JSON format
+      homepage: https://github.com/rpetit3/assembly-scan
+      documentation: https://github.com/rpetit3/assembly-scan
+      tool_dev_url: https://github.com/rpetit3/assembly-scan
+      doi: ""
+      licence: ['MIT']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - assembly:
+      type: file
+      description: FASTA file for a given assembly
+      pattern: "*.fasta"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - json:
+      type: file
+      description: Assembly statistics in JSON format
+      pattern: "*.json"
+
+authors:
+  - "@sateeshperi"
+  - "@mjcipriano"

tests/config/pytest_modules.yml

@@ -30,6 +30,10 @@ artic/minion:
   - modules/artic/minion/**
   - tests/modules/artic/minion/**
 
+assemblyscan:
+  - modules/assemblyscan/**
+  - tests/modules/assemblyscan/**
+
 bamaligncleaner:
   - modules/bamaligncleaner/**
   - tests/modules/bamaligncleaner/**
@@ -1027,14 +1031,14 @@ seacr/callpeak:
   - modules/seacr/callpeak/**
   - tests/modules/seacr/callpeak/**
 
-seqtk/mergepe:
-  - modules/seqtk/mergepe/**
-  - tests/modules/seqtk/mergepe/**
-
 seqkit/split2:
   - modules/seqkit/split2/**
   - tests/modules/seqkit/split2/**
 
+seqtk/mergepe:
+  - modules/seqtk/mergepe/**
+  - tests/modules/seqtk/mergepe/**
+
 seqtk/sample:
   - modules/seqtk/sample/**
   - tests/modules/seqtk/sample/**

tests/modules/assemblyscan/main.nf

@@ -0,0 +1,13 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { ASSEMBLYSCAN } from '../../../modules/assemblyscan/main.nf' addParams( options: [:] )
+
+workflow test_assemblyscan {
+    
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true) ]
+
+    ASSEMBLYSCAN ( input )
+}

tests/modules/assemblyscan/test.yml

@@ -0,0 +1,7 @@
+- name: assemblyscan test_assemblyscan
+  command: nextflow run tests/modules/assemblyscan -entry test_assemblyscan -c tests/config/nextflow.config
+  tags:
+    - assemblyscan
+  files:
+    - path: output/assemblyscan/test.json
+      md5sum: 9140e3d43f2d676f62e1325ace5dd8bd