Add bcftools/query module (#670)

* initial commit [ci skip]

* add basic structure [ci skip]

* finalized the bcftools/query module

* add optional files [ci skip]

* Add the vcf index file as param [ci skip]

* update the md5sum for output file [ci skip]

* all tests passing

modules/bcftools/query/functions.nf

@@ -0,0 +1,68 @@
+//
+//  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"
+        }
+    }
+}

modules/bcftools/query/main.nf

@@ -0,0 +1,49 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process BCFTOOLS_QUERY {
+    tag "$meta.id"
+    label 'process_medium'
+    publishDir "${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::bcftools=1.13" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/bcftools:1.13--h3a49de5_0"
+    } else {
+        container "quay.io/biocontainers/bcftools:1.13--h3a49de5_0"
+    }
+
+    input:
+    tuple val(meta), path(vcf), path(index)
+    path(regions)
+    path(targets)
+    path(samples)
+
+    output:
+    tuple val(meta), path("*.gz") , emit: vcf
+    path "*.version.txt"          , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def regions_file  = regions ? "--regions-file ${regions}" : ""
+    def targets_file = targets ? "--targets-file ${targets}" : ""
+    def samples_file =  samples ? "--samples-file ${samples}" : ""
+
+    """
+    bcftools query \\
+        --output ${prefix}.vcf.gz \\
+        ${regions_file} \\
+        ${targets_file} \\
+        ${samples_file} \\
+        $options.args \\
+        ${vcf}
+
+    echo \$(bcftools --version 2>&1) | sed 's/^.*bcftools //; s/ .*\$//' > ${software}.version.txt
+    """
+}

modules/bcftools/query/meta.yml

@@ -0,0 +1,61 @@
+name: bcftools_query
+description: Extracts fields from VCF or BCF files and outputs them in user-defined format.
+keywords:
+  - query
+  - variant calling
+  - bcftools
+  - VCF
+tools:
+    - query:
+        description: |
+          Extracts fields from VCF or BCF files and outputs them in user-defined format.
+        homepage: http://samtools.github.io/bcftools/bcftools.html
+        documentation: http://www.htslib.org/doc/bcftools.html
+        doi: 10.1093/bioinformatics/btp352
+input:
+    - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information
+          e.g. [ id:'test', single_end:false ]
+    - vcf:
+        type: file
+        description: |
+          The vcf file to be qeuried.
+          e.g. 'file.vcf'
+    - index:
+        type: file
+        description: |
+          The tab index for the VCF file to be inspected.
+          e.g. 'file.tbi'
+    - regions:
+        type: file
+        description: |
+          Optionally, restrict the operation to regions listed in this file.
+          e.g. 'file.vcf'
+    - targets:
+        type: file
+        description: |
+          Optionally, restrict the operation to regions listed in this file (doesn't rely upon index files)
+          e.g. 'file.vcf'
+    - samples:
+        type: file
+        description: |
+          Optional, file of sample names to be included or excluded.
+          e.g. 'file.tsv'
+output:
+    - meta:
+        type: map
+        description: |
+          Groovy Map containing sample information
+          e.g. [ id:'test', single_end:false ]
+    - vcf:
+        type: file
+        description: VCF query output file
+        pattern: "*.{vcf.gz}"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+  - "@abhi18av"

tests/config/pytest_modules.yml

@@ -62,6 +62,10 @@ bcftools/norm:
   - modules/bcftools/norm/**
   - tests/modules/bcftools/norm/**
 
+bcftools/query:
+  - modules/bcftools/query/**
+  - tests/modules/bcftools/query/**
+
 bcftools/reheader:
   - modules/bcftools/reheader/**
   - tests/modules/bcftools/reheader/**

tests/modules/bcftools/query/main.nf

@@ -0,0 +1,31 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { BCFTOOLS_QUERY } from '../../../../modules/bcftools/query/main.nf' addParams( options: ['args': "-f '%CHROM %POS %REF %ALT[%SAMPLE=%GT]'" ] )
+
+workflow test_bcftools_query {
+
+    regions = []
+    targets = []
+    samples = []
+
+    input = [ [ id:'out' ], // meta map
+             file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+             file(params.test_data['sarscov2']['illumina']['test_vcf_gz_tbi'], checkIfExists: true)]
+
+    BCFTOOLS_QUERY ( input, regions, targets, samples )
+}
+
+workflow test_bcftools_query_with_optional_files {
+
+    regions = file(params.test_data['sarscov2']['illumina']['test3_vcf_gz'], checkIfExists: true)
+    targets = file(params.test_data['sarscov2']['illumina']['test2_vcf_targets_tsv_gz'], checkIfExists: true)
+    samples = []
+
+    input = [ [ id:'out' ], // meta map
+             file(params.test_data['sarscov2']['illumina']['test_vcf_gz'], checkIfExists: true),
+             file(params.test_data['sarscov2']['illumina']['test_vcf_gz_tbi'], checkIfExists: true)]
+
+    BCFTOOLS_QUERY ( input, regions, targets, samples )
+}

tests/modules/bcftools/query/test.yml

@@ -0,0 +1,17 @@
+- name: bcftools query
+  command: nextflow run ./tests/modules/bcftools/query -entry test_bcftools_query -c tests/config/nextflow.config
+  tags:
+    - bcftools
+    - bcftools/query
+  files:
+    - path: output/bcftools/out.vcf.gz
+      md5sum: c32a6d28f185822d8fe1eeb7e42ec155
+
+- name: bcftools query with optional files
+  command: nextflow run ./tests/modules/bcftools/query -entry test_bcftools_query_with_optional_files -c tests/config/nextflow.config
+  tags:
+    - bcftools
+    - bcftools/query
+  files:
+    - path: output/bcftools/out.vcf.gz
+      md5sum: 6bb5df49bfb5af39f7037cdf95032aac