add clonalframeml module (#974)

* add clonalframeml module

* Update main.nf

* try recommended gzip

* Update main.nf

Co-authored-by: Chris Cheshire <chris.j.cheshire@gmail.com>

modules/clonalframeml/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+    // Do not publish versions.yml unless running from pytest workflow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/clonalframeml/main.nf

@@ -0,0 +1,47 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process CLONALFRAMEML {
+    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::clonalframeml=1.12" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/clonalframeml:1.12--h7d875b9_1"
+    } else {
+        container "quay.io/biocontainers/clonalframeml:1.12--h7d875b9_1"
+    }
+
+    input:
+    tuple val(meta), path(newick), path(msa)
+
+    output:
+    tuple val(meta), path("*.emsim.txt")                   , emit: emsim, optional: true
+    tuple val(meta), path("*.em.txt")                      , emit: em
+    tuple val(meta), path("*.importation_status.txt")      , emit: status
+    tuple val(meta), path("*.labelled_tree.newick")        , emit: newick
+    tuple val(meta), path("*.ML_sequence.fasta")           , emit: fasta
+    tuple val(meta), path("*.position_cross_reference.txt"), emit: pos_ref
+    path "versions.yml"                                    , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    ClonalFrameML \\
+        $newick \\
+        <(gzip -cdf $msa) \\
+        $prefix \\
+        $options.args
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$( echo \$(ClonalFrameML -version 2>&1) | sed 's/^.*ClonalFrameML v//' )
+    END_VERSIONS
+    """
+}

modules/clonalframeml/meta.yml

@@ -0,0 +1,67 @@
+name: clonalframeml
+description: Predict recomination events in bacterial genomes
+keywords:
+  - fasta
+  - multiple sequence alignment
+  - recombination
+tools:
+  - clonalframeml:
+      description: Efficient inferencing of recombination in bacterial genomes
+      homepage: https://github.com/xavierdidelot/ClonalFrameML
+      documentation: https://github.com/xavierdidelot/clonalframeml/wiki
+      tool_dev_url: https://github.com/xavierdidelot/ClonalFrameML
+      doi: "10.1371/journal.pcbi.1004041"
+      licence: ['GPL v3']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - msa:
+      type: file
+      description: A multiple seqeunce alignmnet in FASTA format
+      pattern: "*.{fasta,fasta.gz,fa,fa.gz,fna,fna.gz}"
+  - newick:
+      type: file
+      description: A Newick formated tree based on multiple sequence alignment
+      pattern: "*.{newick,treefile,dnd}"
+
+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"
+  - emsim:
+      type: file
+      description: Bootstrapped values for the three parameters R/theta, nu and delta
+      pattern: "*.emsim.txt"
+  - em:
+      type: file
+      description: Point estimates for R/theta, nu, delta and the branch lengths
+      pattern: "*.em.txt"
+  - fasta:
+      type: file
+      description: Sequence reconstructed by maximum likelihood
+      pattern: "*.ML_sequence.fasta"
+  - newick:
+      type: file
+      description: Tree with all nodes labelled
+      pattern: "*.labelled_tree.newick"
+  - pos_ref:
+      type: file
+      description: CSV mapping input sequence files to the sequences in the *.ML_sequence.fasta
+      pattern: "*.position_cross_reference.txt"
+  - status:
+      type: file
+      description: List of reconstructed recombination events
+      pattern: "*.importation_status.txt"
+
+authors:
+  - "@rpetit3"

tests/config/pytest_modules.yml

@@ -282,6 +282,10 @@ chromap/index:
   - modules/chromap/index/**
   - tests/modules/chromap/index/**
 
+clonalframeml:
+  - modules/clonalframeml/**
+  - tests/modules/clonalframeml/**
+
 cmseq/polymut:
   - modules/cmseq/polymut/**
   - tests/modules/cmseq/polymut/**
@@ -1210,7 +1214,7 @@ seacr/callpeak:
 seqkit/split2:
   - modules/seqkit/split2/**
   - tests/modules/seqkit/split2/**
-
+ 
 seqsero2:
   - modules/seqsero2/**
   - tests/modules/seqsero2/**

tests/modules/clonalframeml/main.nf

@@ -0,0 +1,14 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { CLONALFRAMEML } from '../../../modules/clonalframeml/main.nf' addParams( options: [:] )
+
+workflow test_clonalframeml {
+    
+    input = [ [ id:'test' ], // meta map
+              file("https://github.com/bactopia/bactopia-tests/raw/main/data/species/haemophilus_influenzae/genome/genome_msa.newick", checkIfExists: true),
+              file("https://github.com/bactopia/bactopia-tests/raw/main/data/species/haemophilus_influenzae/genome/genome_msa.fa.gz", checkIfExists: true),]
+
+    CLONALFRAMEML ( input )
+}

tests/modules/clonalframeml/test.yml

@@ -0,0 +1,15 @@
+- name: clonalframeml test_clonalframeml
+  command: nextflow run tests/modules/clonalframeml -entry test_clonalframeml -c tests/config/nextflow.config
+  tags:
+    - clonalframeml
+  files:
+    - path: output/clonalframeml/test.ML_sequence.fasta
+      md5sum: 1b75cdaea78f5920ebb92125422a2589
+    - path: output/clonalframeml/test.em.txt
+      md5sum: 5439d59897a9a90390bb175207bf2b9b
+    - path: output/clonalframeml/test.importation_status.txt
+      md5sum: 6ce9dbc7746b1c884af042fa02311fba
+    - path: output/clonalframeml/test.labelled_tree.newick
+      md5sum: aa47754eea8a3b6bab56bd7c83ba78db
+    - path: output/clonalframeml/test.position_cross_reference.txt
+      md5sum: 8ff60768b348fc6f7a1e787aca72f596