Bam2fq (#958)

* added template for module

* update main

* added specific code

* wrong variable name in else script

* added tests for both split and nosplit

* docker test successful - updating yaml

* adding echo to version print

modules/samtools/bam2fq/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+    // Do not publish versions.yml unless running from pytest workflow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/samtools/bam2fq/main.nf

@@ -0,0 +1,64 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process SAMTOOLS_BAM2FQ {
+    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::samtools=1.14" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/samtools:1.14--hb421002_0"
+    } else {
+        container "quay.io/biocontainers/samtools:1.14--hb421002_0"
+    }
+
+    input:
+    tuple val(meta), path(inputbam)
+    val(split)
+
+    output:
+    tuple val(meta), path("*.fq.gz"), emit: reads
+    path "versions.yml"          , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+
+    if (split){
+        """
+        samtools \\
+            bam2fq \\
+            $options.args \\
+            -@ $task.cpus \\
+            -1 ${prefix}_1.fq.gz \\
+            -2 ${prefix}_2.fq.gz \\
+            -0 ${prefix}_other.fq.gz \\
+            -s ${prefix}_singleton.fq.gz \\
+            $inputbam
+
+        cat <<-END_VERSIONS > versions.yml
+        ${getProcessName(task.process)}:
+            ${getSoftwareName(task.process)}: \$(echo \$(samtools --version 2>&1) | sed 's/^.*samtools //; s/Using.*\$//')
+        END_VERSIONS
+        """
+    } else {
+        """
+        samtools \\
+            bam2fq \\
+            $options.args \\
+            -@ $task.cpus \\
+            $inputbam >${prefix}_interleaved.fq.gz
+
+        cat <<-END_VERSIONS > versions.yml
+        ${getProcessName(task.process)}:
+            ${getSoftwareName(task.process)}: \$(echo \$(samtools --version 2>&1) | sed 's/^.*samtools //; s/Using.*\$//')
+        END_VERSIONS
+        """
+    }
+
+}

modules/samtools/bam2fq/meta.yml

@@ -0,0 +1,55 @@
+name: samtools_bam2fq
+description: |
+    The module uses bam2fq method from samtools to
+    convert a SAM, BAM or CRAM file to FASTQ format
+keywords:
+  - bam2fq
+  - samtools
+  - fastq
+tools:
+  - samtools:
+      description: Tools for dealing with SAM, BAM and CRAM files
+      homepage: None
+      documentation: http://www.htslib.org/doc/1.1/samtools.html
+      tool_dev_url: None
+      doi: ""
+      licence: ['MIT']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - inputbam:
+      type: file
+      description: BAM/CRAM/SAM file
+      pattern: "*.{bam,cram,sam}"
+  - split:
+      type: boolean
+      description: |
+        TRUE/FALSE value to indicate if reads should be separated into
+        /1, /2 and if present other, or singleton.
+        Note: choosing TRUE will generate 4 different files.
+        Choosing FALSE will produce a single file, which will be interleaved in case
+        the input contains paired reads.
+
+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"
+  - reads:
+      type: file
+      description: |
+        FASTQ files, which will be either a group of 4 files (read_1, read_2, other and singleton)
+        or a single interleaved .fq.gz file if the user chooses not to split the reads.
+      pattern: "*.fq.gz"
+
+authors:
+  - "@lescai"

tests/config/pytest_modules.yml

@@ -983,6 +983,10 @@ samtools/ampliconclip:
   - modules/samtools/ampliconclip/**
   - tests/modules/samtools/ampliconclip/**
 
+samtools/bam2fq:
+  - modules/samtools/bam2fq/**
+  - tests/modules/samtools/bam2fq/**
+
 samtools/depth:
   - modules/samtools/depth/**
   - tests/modules/samtools/depth/**

tests/modules/samtools/bam2fq/main.nf

@@ -0,0 +1,24 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SAMTOOLS_BAM2FQ } from '../../../../modules/samtools/bam2fq/main.nf' addParams( options: [args: "-T RX"] )
+
+workflow test_samtools_bam2fq_nosplit {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_paired_end_umi_converted_bam'], checkIfExists: true) ]
+    split = false
+
+    SAMTOOLS_BAM2FQ ( input, split )
+}
+
+
+workflow test_samtools_bam2fq_withsplit {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_paired_end_umi_converted_bam'], checkIfExists: true) ]
+    split = true
+
+    SAMTOOLS_BAM2FQ ( input, split )
+}

tests/modules/samtools/bam2fq/test.yml

@@ -0,0 +1,23 @@
+- name: samtools bam2fq test_samtools_bam2fq_nosplit
+  command: nextflow run tests/modules/samtools/bam2fq -entry test_samtools_bam2fq_nosplit -c tests/config/nextflow.config
+  tags:
+    - samtools/bam2fq
+    - samtools
+  files:
+    - path: output/samtools/test_interleaved.fq.gz
+      md5sum: d733e66d29a4b366bf9df8c42f845256
+
+- name: samtools bam2fq test_samtools_bam2fq_withsplit
+  command: nextflow run tests/modules/samtools/bam2fq -entry test_samtools_bam2fq_withsplit -c tests/config/nextflow.config
+  tags:
+    - samtools/bam2fq
+    - samtools
+  files:
+    - path: output/samtools/test_1.fq.gz
+      md5sum: 4522edbe158ec4804765794569f67493
+    - path: output/samtools/test_2.fq.gz
+      md5sum: 7e00ef40d5cfe272b67461381019dcc1
+    - path: output/samtools/test_other.fq.gz
+      md5sum: 709872fc2910431b1e8b7074bfe38c67
+    - path: output/samtools/test_singleton.fq.gz
+      md5sum: 709872fc2910431b1e8b7074bfe38c67