Merge pull request #168 from JoseEspinosa/bcftools_merge

Add bcftools merge module

.github/filters.yml

@@ -18,6 +18,10 @@ bcftools_isec:
   - software/bcftools/isec/**
   - tests/software/bcftools/isec/**
 
+bcftools_merge:
+  - software/bcftools/merge/**
+  - tests/software/bcftools/merge/**
+
 bcftools_stats:
   - software/bcftools/stats/**
   - tests/software/bcftools/stats/**

software/bcftools/merge/functions.nf

@@ -0,0 +1,60 @@
+
+/*
+ * -----------------------------------------------------
+ *  Utility functions used in nf-core DSL2 module files
+ * -----------------------------------------------------
+ */
+
+/*
+ * Extract name of software tool from process name using $task.process
+ */
+def getSoftwareName(task_process) {
+    return task_process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()
+}
+
+/*
+ * Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+ */
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args          = args.args ?: ''
+    options.args2         = args.args2 ?: ''
+    options.publish_by_id = args.publish_by_id ?: false
+    options.publish_dir   = args.publish_dir ?: ''
+    options.publish_files = args.publish_files
+    options.suffix        = args.suffix ?: ''
+    return options
+}
+
+/*
+ * Tidy up and join elements of a list to return a path string
+ */
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }  // Remove empty entries
+    paths = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", '') } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+/*
+ * Function to save/publish module results
+ */
+def saveFiles(Map args) {
+    if (!args.filename.endsWith('.version.txt')) {
+        def ioptions = initOptions(args.options)
+        def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+        if (ioptions.publish_by_id) {
+            path_list.add(args.publish_id)
+        }
+        if (ioptions.publish_files instanceof Map) {
+            for (ext in ioptions.publish_files) {
+                if (args.filename.endsWith(ext.key)) {
+                    def ext_list = path_list.collect()
+                    ext_list.add(ext.value)
+                    return "${getPathFromList(ext_list)}/$args.filename"
+                }
+            }
+        } else if (ioptions.publish_files == null) {
+            return "${getPathFromList(path_list)}/$args.filename"
+        }
+    }
+}

software/bcftools/merge/main.nf

@@ -0,0 +1,37 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+def options    = initOptions(params.options)
+
+process BCFTOOLS_MERGE {
+    tag "$meta.id"
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), publish_id:meta.id) }
+
+    conda (params.enable_conda ? "bioconda::bcftools=1.11=h7c999a4_0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/bcftools:1.11--h7c999a4_0"
+    } else {
+        container "quay.io/biocontainers/bcftools:1.11--h7c999a4_0"
+    }
+
+    input:
+    tuple val(meta), path(vcfs), path(tbis)
+
+    output:
+    tuple val(meta), path("*.gz"), emit: vcf
+    path  "*.version.txt"        , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    prefix       = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    bcftools merge -Oz \\
+        --output ${prefix}.vcf.gz \\
+        $options.args \\
+        *.vcf.gz
+    echo \$(bcftools --version 2>&1) | sed 's/^.*bcftools //; s/ .*\$//' > ${software}.version.txt
+    """
+}

software/bcftools/merge/meta.yml

@@ -0,0 +1,67 @@
+name: bcftools_merge
+description: Merge VCF files
+keywords:
+    - variant calling
+    - merge
+    - VCF
+tools:
+    - isec:
+        description: |
+          Merge VCF files.
+        homepage: http://samtools.github.io/bcftools/bcftools.html
+        documentation: http://www.htslib.org/doc/bcftools.html
+        doi: 10.1093/bioinformatics/btp352
+params:
+    - outdir:
+        type: string
+        description: |
+          The pipeline's output directory. By default, the module will
+          output files into `$params.outdir/<SOFTWARE>`
+    - publish_dir_mode:
+        type: string
+        description: |
+          Value for the Nextflow `publishDir` mode parameter.
+          Available: symlink, rellink, link, copy, copyNoFollow, move.
+    - enable_conda:
+        type: boolean
+        description: |
+          Run the module with Conda using the software specified
+          via the `conda` directive
+    - singularity_pull_docker_container:
+        type: boolean
+        description: |
+          Instead of directly downloading Singularity images for use with Singularity,
+          force the workflow to pull and convert Docker containers instead.
+input:
+    - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information
+          e.g. [ id:'test', single_end:false ]
+    - vcfs:
+        type: files
+        description: |
+          List containing 2 or more vcf files
+          e.g. [ 'file1.vcf', 'file2.vcf' ]
+    - tbis:
+        type: files
+        description: |
+          List containing the tbi index files corresponding to the vcfs input files
+          e.g. [ 'file1.vcf.tbi', 'file2.vcf.tbi' ]
+output:
+    - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information
+          e.g. [ id:'test', single_end:false ]
+    - vcf:
+        type: file
+        description: VCF merged output file
+        pattern: "*.{vcf.gz}"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@joseespinosa"
+    - "@drpatelh"

tests/software/bcftools/merge/main.nf

@@ -0,0 +1,18 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+//keep --no-verson argument, otherwise md5 will change on each execution
+include { BCFTOOLS_MERGE } from '../../../../software/bcftools/merge/main.nf' addParams( options: ['args': '--force-samples --no-version'] )
+
+workflow test_bcftools_merge {
+
+    def input = []
+    input = [ [ id:'test' ], // meta map
+              [ file("${launchDir}/tests/data/vcf/test_a.merge.vcf.gz", checkIfExists: true),
+                file("${launchDir}/tests/data/vcf/test_b.merge.vcf.gz", checkIfExists: true) ],
+              [ file("${launchDir}/tests/data/vcf/test_a.merge.vcf.gz.tbi", checkIfExists: true),
+                file("${launchDir}/tests/data/vcf/test_b.merge.vcf.gz.tbi", checkIfExists: true) ]]
+
+    BCFTOOLS_MERGE ( input )
+}

tests/software/bcftools/merge/test.yml

@@ -0,0 +1,8 @@
+- name: bcftools merge
+  command: nextflow run ./tests/software/bcftools/merge -entry test_bcftools_merge -c tests/config/nextflow.config
+  tags:
+    - bcftools
+    - bcftools_merge
+  files:
+    - path: output/bcftools/test.vcf.gz
+      md5sum: 23eee1acfd99581889be3fd2b86b6950