add module for ariba (#1731)

2 years ago · 1fe4cb942f
parent 30f72e2482
commit 1fe4cb942f
11 changed files with 228 additions and 0 deletions
--- a/modules/ariba/getref/main.nf
+++ b/modules/ariba/getref/main.nf
@ -0,0 +1,41 @@
+process ARIBA_GETREF {
+    tag "$db_name"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::ariba=2.14.6" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/ariba:2.14.6--py39h67e14b5_3':
+        'quay.io/biocontainers/ariba:2.14.6--py39h67e14b5_3' }"
+
+    input:
+    val(db_name)
+
+    output:
+    tuple path("${db_name}.tar.gz"), emit: db
+    path "versions.yml"            , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    """
+    # Download, format database, and tarball it
+    ariba \\
+        getref \\
+        ${db_name} \\
+        ${db_name}
+
+    ariba \\
+        prepareref \\
+        -f ${db_name}.fa \\
+        -m ${db_name}.tsv \\
+        ${db_name}
+
+    tar -zcvf ${db_name}.tar.gz ${db_name}/
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ariba:  \$(echo \$(ariba version 2>&1) | sed 's/^.*ARIBA version: //;s/ .*\$//')
+    END_VERSIONS
+    """
+}
--- a/modules/ariba/getref/meta.yml
+++ b/modules/ariba/getref/meta.yml
@ -0,0 +1,33 @@
+name: "ariba_getref"
+description: Download and prepare database for Ariba analysis
+keywords:
+  - fastq
+  - assembly
+  - resistance
+  - virulence
+tools:
+  - "ariba":
+      description: "ARIBA: Antibiotic Resistance Identification By Assembly"
+      homepage: "https://sanger-pathogens.github.io/ariba/"
+      documentation: "https://sanger-pathogens.github.io/ariba/"
+      tool_dev_url: "https://github.com/sanger-pathogens/ariba"
+      doi: "110.1099/mgen.0.000131"
+      licence: "['GPL v3']"
+
+input:
+  - db_name:
+      type: string
+      description: A database to setup up for Ariba
+
+output:
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - db:
+      type: file
+      description: An Ariba prepared database
+      pattern: "*.tar.gz"
+
+authors:
+  - "@rpetit3"
--- a/modules/ariba/run/main.nf
+++ b/modules/ariba/run/main.nf
@ -0,0 +1,40 @@
+process ARIBA_RUN {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::ariba=2.14.6" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/ariba:2.14.6--py39h67e14b5_3':
+        'quay.io/biocontainers/ariba:2.14.6--py39h67e14b5_3' }"
+
+    input:
+    tuple val(meta), path(reads)
+    each path(db)
+
+    output:
+    tuple val(meta), path("${prefix}/*"), emit: results
+    path "versions.yml"                 , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def db_name = db.getName().replace('.tar.gz', '')
+    """
+    tar -xzvf ${db}
+    ariba \\
+        run \\
+        ${db_name}/ \\
+        ${reads} \\
+        ${prefix} \\
+        $args \\
+        --threads $task.cpus
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        ariba:  \$(echo \$(ariba version 2>&1) | sed 's/^.*ARIBA version: //;s/ .*\$//')
+    END_VERSIONS
+    """
+}
--- a/modules/ariba/run/meta.yml
+++ b/modules/ariba/run/meta.yml
@ -0,0 +1,48 @@
+name: "ariba_run"
+description: Query input FASTQs against Ariba formatted databases
+keywords:
+  - fastq
+  - assembly
+  - resistance
+  - virulence
+tools:
+  - "ariba":
+      description: "ARIBA: Antibiotic Resistance Identification By Assembly"
+      homepage: "https://sanger-pathogens.github.io/ariba/"
+      documentation: "https://sanger-pathogens.github.io/ariba/"
+      tool_dev_url: "https://github.com/sanger-pathogens/ariba"
+      doi: "110.1099/mgen.0.000131"
+      licence: "['GPL v3']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - reads:
+      type: file
+      description: Paired-end reads in FASTQ format
+      pattern: "*_R[1|2].{fastq.gz,fq.gz}"
+  - db:
+      type: file
+      description: An Ariba prepared database
+      pattern: "*.tar.gz"
+
+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"
+  - results:
+      type: file
+      description: A directory of Ariba analysis outputs
+      pattern: "*"
+
+authors:
+  - "@rpetit3"
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@ -54,6 +54,14 @@ antismash/antismashlitedownloaddatabases:
  - modules/antismash/antismashlitedownloaddatabases/**
  - tests/modules/antismash/antismashlitedownloaddatabases/**

+ariba/getref:
+  - modules/ariba/getref/**
+  - tests/modules/ariba/getref/**
+
+ariba/run:
+  - modules/ariba/run/**
+  - tests/modules/ariba/run/**
+
 arriba:
  - modules/arriba/**
  - tests/modules/arriba/**
--- a/tests/modules/ariba/getref/main.nf
+++ b/tests/modules/ariba/getref/main.nf
@ -0,0 +1,9 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { ARIBA_GETREF } from '../../../../modules/ariba/getref/main.nf'
+
+workflow test_ariba_getref {
+    ARIBA_GETREF ( "card" )
+}
--- a/tests/modules/ariba/getref/nextflow.config
+++ b/tests/modules/ariba/getref/nextflow.config
@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    
+}
--- a/tests/modules/ariba/getref/test.yml
+++ b/tests/modules/ariba/getref/test.yml
@ -0,0 +1,7 @@
+- name: ariba getref test_ariba_getref
+  command: nextflow run ./tests/modules/ariba/getref -entry test_ariba_getref -c ./tests/config/nextflow.config  -c ./tests/modules/ariba/getref/nextflow.config
+  tags:
+    - ariba
+    - ariba/getref
+  files:
+    - path: output/ariba/card.tar.gz
--- a/tests/modules/ariba/run/main.nf
+++ b/tests/modules/ariba/run/main.nf
@ -0,0 +1,17 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { ARIBA_GETREF } from '../../../../modules/ariba/getref/main.nf'
+include { ARIBA_RUN } from '../../../../modules/ariba/run/main.nf'
+
+workflow test_ariba_run {
+    
+    input = [ [ id:'test', single_end:false ], // meta map
+              [  file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+                 file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) ]
+            ]
+
+    ARIBA_GETREF ( "card" )
+    ARIBA_RUN ( input, ARIBA_GETREF.out.db)
+}
--- a/tests/modules/ariba/run/nextflow.config
+++ b/tests/modules/ariba/run/nextflow.config
@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    
+}
--- a/tests/modules/ariba/run/test.yml
+++ b/tests/modules/ariba/run/test.yml
@ -0,0 +1,15 @@
+- name: ariba run test_ariba_run
+  command: nextflow run ./tests/modules/ariba/run -entry test_ariba_run -c ./tests/config/nextflow.config  -c ./tests/modules/ariba/run/nextflow.config
+  tags:
+    - ariba
+    - ariba/run
+  files:
+    - path: output/ariba/test/assembled_genes.fa.gz
+    - path: output/ariba/test/assembled_seqs.fa.gz
+    - path: output/ariba/test/assemblies.fa.gz
+    - path: output/ariba/test/debug.report.tsv
+      contains: ["ariba_ref_name", "reads", "free_text", "known_var_change"]
+    - path: output/ariba/test/report.tsv
+      contains: ["ariba_ref_name", "reads", "free_text", "known_var_change"]
+    - path: output/ariba/test/version_info.txt
+      contains: ["ARIBA", "Python", "OK", "True"]