add hicap module (#772)

* add hicap module

* add info on optional inputs

* fix typo

* Update meta.yml

* Update main.nf

* Update meta.yml

* Update modules/hicap/main.nf

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

* Update modules/hicap/main.nf

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

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+    // Do not publish versions.yml unless running from pytest workflow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

@@ -0,0 +1,56 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process HICAP {
+    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::hicap=1.0.3" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/hicap:1.0.3--py_0"
+    } else {
+        container "quay.io/biocontainers/hicap:1.0.3--py_0"
+    }
+
+    input:
+    tuple val(meta), path(fasta)
+    path database_dir
+    path model_fp
+
+    output:
+    tuple val(meta), path("*.gbk"), emit: gbk
+    tuple val(meta), path("*.svg"), emit: svg
+    tuple val(meta), path("*.tsv"), emit: tsv
+    path "versions.yml"           , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def database_args = database_dir ? "--database_dir ${database_dir}" : ""
+    def model_args = model_fp ? "--model_fp ${model_fp}" : ""
+    def is_compressed = fasta.getName().endsWith(".gz") ? true : false
+    def fasta_name = fasta.getName().replace(".gz", "")
+    """
+    if [ "$is_compressed" == "true" ]; then
+        gzip -c -d $fasta > $fasta_name
+    fi
+
+    hicap \\
+        --query_fp $fasta_name \\
+        $database_args \\
+        $model_args \\
+        $options.args \\
+        --threads $task.cpus \\
+        -o ./
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$( echo \$( hicap --version 2>&1 ) | sed 's/^.*hicap //' )
+    END_VERSIONS
+    """
+}

modules/hicap/meta.yml

@@ -0,0 +1,59 @@
+name: hicap
+description: Identify cap locus serotype and structure in your Haemophilus influenzae assemblies
+keywords:
+  - fasta
+  - serotype
+  - Haemophilus influenzae
+tools:
+  - hicap:
+      description: In silico typing of the H. influenzae capsule locus
+      homepage: https://github.com/scwatts/hicap
+      documentation: https://github.com/scwatts/hicap
+      tool_dev_url: https://github.com/scwatts/hicap
+      doi: "https://doi.org/10.1128/JCM.00190-19"
+      licence: ['GPL v3']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - fasta:
+      type: file
+      description: FASTA formatted assembly file
+      pattern: "*.{fasta,fasta.gz,fa,fa.gz,fna,fna.gz}"
+  - database_dir:
+      type: directory
+      description: Optional - Directory containing locus database
+      pattern: "*/*"
+  - model_fp:
+      type: file
+      description: Optional - Prodigal model to use for gene prediction
+      pattern: "*.{bin}"
+
+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"
+  - gbk:
+      type: file
+      description: GenBank file and cap locus annotations
+      pattern: "*.gbk"
+  - svg:
+      type: file
+      description: Visualization of annotated cap locus
+      pattern: "*.svg"
+  - tsv:
+      type: file
+      description: Detailed summary of cap locus annotations
+      pattern: "*.tsv"
+
+authors:
+  - "@rpetit3"

tests/config/pytest_modules.yml

@@ -470,6 +470,10 @@ gunzip:
   - modules/gunzip/**
   - tests/modules/gunzip/**
 
+hicap:
+  - modules/hicap/**
+  - tests/modules/hicap/**
+
 hifiasm:
   - modules/hifiasm/**
   - tests/modules/hifiasm/**

tests/modules/hicap/main.nf

@@ -0,0 +1,16 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { HICAP } from '../../../modules/hicap/main.nf' addParams( options: [:] )
+
+workflow test_hicap {
+    
+    input = [ [ id:'test', single_end:false ], // meta map
+              file("https://github.com/bactopia/bactopia-tests/raw/main/data/species-specific/haemophilus-influenzae/GCF_900478275.fna.gz", checkIfExists: true) ]
+    
+    database_dir = []
+    model_fp = []
+
+    HICAP ( input, database_dir, model_fp )
+}

tests/modules/hicap/test.yml

@@ -0,0 +1,10 @@
+- name: hicap test_hicap
+  command: nextflow run tests/modules/hicap -entry test_hicap -c tests/config/nextflow.config
+  tags:
+    - hicap
+  files:
+    - path: output/hicap/GCF_900478275.gbk
+      md5sum: 562d026956903354ac80721f501335d4
+    - path: output/hicap/GCF_900478275.svg
+      md5sum: 4fb94871dd0fdd8b4496049668176631
+    - path: output/hicap/GCF_900478275.tsv