Add Nextclade module (#484)

* Bump pangolin version

* Add nextclade to software list

* Add nextclade module

* Update md5sum for Pangolin due to version bump

* Adding some URL to meta.yml

* Adding new line at end of file

Co-authored-by: JoseEspinosa <kadomu@gmail.com>

software/nextclade/functions.nf

@@ -0,0 +1,70 @@
+/*
+ * -----------------------------------------------------
+ *  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"
+        }
+    }
+}

software/nextclade/main.nf

@@ -0,0 +1,55 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process NEXTCLADE {
+    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::nextclade_js=0.14.2" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/nextclade_js:0.14.2--h9ee0642_0"
+    } else {
+        container "quay.io/biocontainers/nextclade_js:0.14.2--h9ee0642_0"
+    }
+
+    input:
+    tuple val(meta), path(fasta)
+    val   output_format
+
+    output:
+    tuple val(meta), path("${prefix}.csv")       , optional:true, emit: csv
+    tuple val(meta), path("${prefix}.json")      , optional:true, emit: json
+    tuple val(meta), path("${prefix}.tree.json") , optional:true, emit: json_tree
+    tuple val(meta), path("${prefix}.tsv")       , optional:true, emit: tsv
+    tuple val(meta), path("${prefix}.clades.tsv"), optional:true, emit: tsv_clades
+    path "*.version.txt"                         , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    prefix       = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def format   = output_format
+    if (!(format in ['json', 'csv', 'tsv', 'tree', 'tsv-clades-only'])) {
+        format = 'json'
+    }
+    def extension = format
+    if (format in ['tsv-clades-only']) {
+        extension = '.clades.tsv'
+    } else if (format in ['tree']) {
+        extension = 'tree.json'
+    }
+    """
+    nextclade \\
+        $options.args \\
+        --jobs $task.cpus \\
+        --input-fasta $fasta \\
+        --output-${format} ${prefix}.${extension}
+
+    echo \$(nextclade --version 2>&1) > ${software}.version.txt
+    """
+}

software/nextclade/meta.yml

@@ -0,0 +1,64 @@
+name: nextclade
+description: SARS-CoV-2 genome clade assignment, mutation calling, and sequence quality checks (Javascript implementation)
+keywords:
+  - nextclade
+  - variant
+  - consensus
+tools:
+  - nextclade:
+      description: SARS-CoV-2 genome clade assignment, mutation calling, and sequence quality checks (Javascript implementation)
+      homepage: https://clades.nextstrain.org
+      documentation: None
+      tool_dev_url: https://github.com/nextstrain/nextclade
+      doi: ""
+      licence: ['MIT']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - fasta:
+      type: file
+      description: FASTA file containing one or more consensus sequences
+      pattern: "*.{fasta,fa}"
+  - output_format:
+      type: string
+      description: |
+        String for output format supported by nextclade
+        i.e one of 'json', 'csv', 'tsv', 'tree', 'tsv-clades-only'
+
+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}"
+  - csv:
+      type: file
+      description: CSV file containing nextclade results
+      pattern: "*.{csv}"
+  - json:
+      type: file
+      description: JSON file containing nextclade results
+      pattern: "*.{json}"
+  - json_tree:
+      type: file
+      description: Auspice JSON V2 containing nextclade results
+      pattern: "*.{tree.json}"
+  - tsv:
+      type: file
+      description: TSV file containing nextclade results
+      pattern: "*.{tsv}"
+  - tsv_clades:
+      type: file
+      description: TSV file containing nextclade results for clades only
+      pattern: "*.{clades.tsv}"
+
+authors:
+  - "@drpatelh"

software/pangolin/main.nf

@@ -11,11 +11,11 @@ process PANGOLIN {
         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::pangolin=2.3.8' : null)
+    conda (params.enable_conda ? 'bioconda::pangolin=2.4.2' : null)
     if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
-        container 'https://depot.galaxyproject.org/singularity/pangolin:2.3.8--py_0'
+        container 'https://depot.galaxyproject.org/singularity/pangolin:2.4.2--pyhdfd78af_0'
     } else {
-        container 'quay.io/biocontainers/pangolin:2.3.8--py_0'
+        container 'quay.io/biocontainers/pangolin:2.4.2--pyhdfd78af_0'
     }
 
     input:

tests/config/pytest_software.yml

@@ -411,6 +411,10 @@ nanoplot:
   - software/nanoplot/**
   - tests/software/nanoplot/**
 
+nextclade:
+  - software/nextclade/**
+  - tests/software/nextclade/**
+
 optitype:
   - software/optitype/**
   - tests/software/optitype/**

tests/software/nextclade/main.nf

@@ -0,0 +1,33 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { NEXTCLADE } from '../../../software/nextclade/main.nf' addParams( options: [:] )
+
+workflow test_nextclade_json {
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    NEXTCLADE ( input, 'json' )
+}
+
+workflow test_nextclade_csv {
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    NEXTCLADE ( input, 'csv' )
+}
+
+workflow test_nextclade_tsv {
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    NEXTCLADE ( input, 'tsv' )
+}
+
+workflow test_nextclade_tree {
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+            ]
+    NEXTCLADE ( input, 'tree' )
+}

tests/software/nextclade/test.yml

@@ -0,0 +1,31 @@
+- name: nextclade test_nextclade_json
+  command: nextflow run tests/software/nextclade -entry test_nextclade_json -c tests/config/nextflow.config
+  tags:
+    - nextclade
+  files:
+    - path: output/nextclade/test.json
+      md5sum: cab92830c5cb66076e7d6c054ea98362
+
+- name: nextclade test_nextclade_csv
+  command: nextflow run tests/software/nextclade -entry test_nextclade_csv -c tests/config/nextflow.config
+  tags:
+    - nextclade
+  files:
+    - path: output/nextclade/test.csv
+      md5sum: 4f7096df9be51f99a0d62a38653b29cf
+
+- name: nextclade test_nextclade_tsv
+  command: nextflow run tests/software/nextclade -entry test_nextclade_tsv -c tests/config/nextflow.config
+  tags:
+    - nextclade
+  files:
+    - path: output/nextclade/test.tsv
+      md5sum: fe07dc4ffcd81742ca9bef93f88e8836
+
+- name: nextclade test_nextclade_tree
+  command: nextflow run tests/software/nextclade -entry test_nextclade_tree -c tests/config/nextflow.config
+  tags:
+    - nextclade
+  files:
+    - path: output/nextclade/test.tree.json
+      md5sum: 508a6ea19daf98fd8ae9f2d470f8c3cd

tests/software/pangolin/test.yml

@@ -4,4 +4,4 @@
     - pangolin
   files:
     - path: ./output/pangolin/test.pangolin.csv
-      md5sum: c8b1720f98c9e032908f61bbc05a0fe2
+      md5sum: 57f6e1df5caa2515b0f1cefedeed6770