add prokka module (#298)

* add prokka module

* adjust test outputs

* fix prokka command

* adjust test outputs

* handle optional input

* update pytest

* update pytest

* adjust test outputs

* adjust test outputs

* switch to lists for optional inputs, add md5sums for non-timestamped outputs

* add optional prodigal training file

* add optional prodigal training info to meta yaml

software/prokka/functions.nf

@@ -0,0 +1,60 @@
+/*
+ * -----------------------------------------------------
+ *  Utility functions used in nf-core DSL2 module files
+ * -----------------------------------------------------
+ */
+
+/*
+ * Extract name of software tool from process name using $task.process
+ */
+def getSoftwareName(task_process) {
+    return task_process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()
+}
+
+/*
+ * Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+ */
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args          = args.args ?: ''
+    options.args2         = args.args2 ?: ''
+    options.args3         = args.args3 ?: ''
+    options.publish_by_id = args.publish_by_id ?: false
+    options.publish_dir   = args.publish_dir ?: ''
+    options.publish_files = args.publish_files
+    options.suffix        = args.suffix ?: ''
+    return options
+}
+
+/*
+ * Tidy up and join elements of a list to return a path string
+ */
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }  // Remove empty entries
+    paths = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", "") } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+/*
+ * Function to save/publish module results
+ */
+def saveFiles(Map args) {
+    if (!args.filename.endsWith('.version.txt')) {
+        def ioptions = initOptions(args.options)
+        def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+        if (ioptions.publish_by_id) {
+            path_list.add(args.publish_id)
+        }
+        if (ioptions.publish_files instanceof Map) {
+            for (ext in ioptions.publish_files) {
+                if (args.filename.endsWith(ext.key)) {
+                    def ext_list = path_list.collect()
+                    ext_list.add(ext.value)
+                    return "${getPathFromList(ext_list)}/$args.filename"
+                }
+            }
+        } else if (ioptions.publish_files == null) {
+            return "${getPathFromList(path_list)}/$args.filename"
+        }
+    }
+}

software/prokka/main.nf

@@ -0,0 +1,56 @@
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process PROKKA {
+    tag "$meta.id"
+    label 'prokka'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), publish_id:meta.id) }
+
+    conda (params.enable_conda ? "bioconda::prokka=1.14.6" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/prokka:1.14.6--pl526_0"
+    } else {
+        container "quay.io/biocontainers/prokka:1.14.6--pl526_0"
+    }
+
+    input:
+    tuple val(meta), path(fasta)
+    path proteins
+    path prodigal_tf
+
+    output:
+    tuple val(meta), path("${prefix}/*.gff"), emit: gff
+    tuple val(meta), path("${prefix}/*.gbk"), emit: gbk
+    tuple val(meta), path("${prefix}/*.fna"), emit: fna
+    tuple val(meta), path("${prefix}/*.faa"), emit: faa
+    tuple val(meta), path("${prefix}/*.ffn"), emit: ffn
+    tuple val(meta), path("${prefix}/*.sqn"), emit: sqn
+    tuple val(meta), path("${prefix}/*.fsa"), emit: fsa
+    tuple val(meta), path("${prefix}/*.tbl"), emit: tbl
+    tuple val(meta), path("${prefix}/*.err"), emit: err
+    tuple val(meta), path("${prefix}/*.log"), emit: log
+    tuple val(meta), path("${prefix}/*.txt"), emit: txt
+    tuple val(meta), path("${prefix}/*.tsv"), emit: tsv
+    path "*.version.txt", emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def proteins_opt = proteins ? "--proteins ${proteins[0]}" : ""
+    def prodigal_opt = prodigal_tf ? "--prodigaltf ${prodigal_tf[0]}" : ""
+    """
+    prokka \\
+        $options.args \\
+        --cpus $task.cpus \\
+        --prefix $prefix \\
+        $proteins_opt \\
+        $prodigal_tf \\
+        $fasta
+
+    echo \$(prokka --version 2>&1) | sed 's/^.*prokka //' > ${software}.version.txt
+    """
+}

software/prokka/meta.yml

@@ -0,0 +1,91 @@
+name: prokka
+description: Whole genome annotation of small genomes (bacterial, archeal, viral)
+keywords:
+  - annotation
+  - fasta
+  - prokka
+tools:
+  - prokka:
+      description: Rapid annotation of prokaryotic genomes
+      homepage: https://github.com/tseemann/prokka
+      doi: "10.1093/bioinformatics/btu153"
+      licence: ['GPL v2']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - fasta:
+      type: file
+      description: |
+        FASTA file to be annotated. Has to contain at least a non-empty string dummy value.
+  - proteins:
+      type: file
+      description: FASTA file of trusted proteins to first annotate from (optional)
+  - prodigal_tf:
+      type: file
+      description: Training file to use for Prodigal (optional)
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "*.{version.txt}"
+  - gff:
+      type: file
+      description: annotation in GFF3 format, containing both sequences and annotations
+      pattern: "*.{gff}"
+  - gbk:
+      type: file
+      description: annotation in GenBank format, containing both sequences and annotations
+      pattern: "*.{gbk}"
+  - fna:
+      type: file
+      description: nucleotide FASTA file of the input contig sequences
+      pattern: "*.{fna}"
+  - faa:
+      type: file
+      description: protein FASTA file of the translated CDS sequences
+      pattern: "*.{faa}"
+  - ffn:
+      type: file
+      description: nucleotide FASTA file of all the prediction transcripts (CDS, rRNA, tRNA, tmRNA, misc_RNA)
+      pattern: "*.{ffn}"
+  - sqn:
+      type: file
+      description: an ASN1 format "Sequin" file for submission to Genbank
+      pattern: "*.{sqn}"
+  - fsa:
+      type: file
+      description: nucleotide FASTA file of the input contig sequences, used by "tbl2asn" to create the .sqn file
+      pattern: "*.{fsa}"
+  - tbl:
+      type: file
+      description: feature Table file, used by "tbl2asn" to create the .sqn file
+      pattern: "*.{tbl}"
+  - err:
+      type: file
+      description: unacceptable annotations - the NCBI discrepancy report.
+      pattern: "*.{err}"
+  - log:
+      type: file
+      description: contains all the output that Prokka produced during its run
+      pattern: "*.{log}"
+  - txt:
+      type: file
+      description: statistics relating to the annotated features found
+      pattern: "*.{txt}"
+  - tsv:
+      type: file
+      description: tab-separated file of all features (locus_tag,ftype,len_bp,gene,EC_number,COG,product)
+      pattern: "*.{tsv}"
+
+authors:
+  - "@rpetit3"

tests/config/pytest_software.yml

@@ -283,6 +283,10 @@ preseq_lcextrap:
   - software/preseq/lcextrap/**
   - tests/software/preseq/lcextrap/**
 
+prokka:
+  - software/prokka/**
+  - tests/software/prokka/**
+
 qualimap_bamqc:
   - software/qualimap/bamqc/**
   - tests/software/qualimap/bamqc/**

tests/software/prokka/main.nf

@@ -0,0 +1,12 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { PROKKA } from '../../../software/prokka/main.nf' addParams( options: [:] )
+
+workflow test_prokka {
+    def input = []
+    input = [ [ id:'test', single_end:false ], // meta map
+              file("${launchDir}/tests/data/genomics/sarscov2/fasta/test_genome.fasta", checkIfExists: true), ]
+    PROKKA ( input, [], [] )
+}

tests/software/prokka/test.yml

@@ -0,0 +1,27 @@
+- name: prokka
+  command: nextflow run ./tests/software/prokka -entry test_prokka -c tests/config/nextflow.config
+  tags:
+    - prokka
+  files:
+    - path: output/prokka/test/test.err
+      md5sum: b3daedc646fddd422824e2b3e5e9229d
+    - path: output/prokka/test/test.faa
+      md5sum: a4ceda83262b3c222a6b1f508fb9e24b
+    - path: output/prokka/test/test.ffn
+      md5sum: 80f474b5367b7ea5ed23791935f65e34
+    - path: output/prokka/test/test.fna
+      md5sum: 787307f29a263e5657cc276ebbf7e2b3
+    - path: output/prokka/test/test.fsa
+      md5sum: 71bbefcb7f12046bcd3263f58cfd5404
+    - path: output/prokka/test/test.gff
+      md5sum: 5dbfb8fcf2db020564c16045976a0933
+    - path: output/prokka/test/test.tbl
+      md5sum: d8f816a066ced94b62d9618b13fb8add
+    - path: output/prokka/test/test.tsv
+      md5sum: da7c720c3018c5081d6a70b517b7d450
+    - path: output/prokka/test/test.txt
+      md5sum: b40e485ffc8eaf1feacf8d79d9751a33
+    # Contain time stamps
+    - path: output/prokka/test/test.gbk
+    - path: output/prokka/test/test.log
+    - path: output/prokka/test/test.sqn