New modules added: issues #200 and #310 (#884)

* New modules added: issues #200 and #310

* Update main.nf

* Update meta.yml

* Update tests/modules/gatk4/genotypegvcfs/main.nf

* Apply suggestions from code review

* Update main.nf

* Updating tests for GenomicsDB input and adding the path for this test resource to test_data.config

* Some minor changes on one of the test files I forgot to include

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>
Co-authored-by: GCJMackenzie <43276267+GCJMackenzie@users.noreply.github.com>

modules/gatk4/genotypegvcfs/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+    // Do not publish versions.yml unless running from pytest workflow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/gatk4/genotypegvcfs/main.nf

@@ -0,0 +1,54 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process GATK4_GENOTYPEGVCFS {
+    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::gatk4=4.2.0.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/gatk4:4.2.0.0--0"
+    } else {
+        container "quay.io/biocontainers/gatk4:4.2.0.0--0"
+    }
+
+    input:
+    tuple val(meta), path(gvcf), path(gvcf_index)
+    path  fasta
+    path  fasta_index
+    path  fasta_dict
+    path  dbsnp
+    path  dbsnp_index
+    path  intervals_bed
+
+    output:
+    tuple val(meta), path("*.vcf.gz"), emit: vcf
+    path  "versions.yml"             , emit: versions
+
+    script:
+    def prefix           = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def dbsnp_options    = dbsnp ? "-D ${dbsnp}" : ""
+    def interval_options = intervals_bed ? "-L ${intervals_bed}" : ""
+    def gvcf_options     = gvcf.name.endsWith(".vcf") || gvcf.name.endsWith(".vcf.gz") ? "$gvcf" : "gendb://$gvcf"
+    """
+    gatk \\
+        GenotypeGVCFs \\
+        $options.args \\
+        $interval_options \\
+        $dbsnp_options \\
+        -R $fasta \\
+        -V $gvcf_options \\
+        -O ${prefix}.vcf.gz
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$(echo \$(gatk --version 2>&1) | sed 's/^.*(GATK) v//; s/ .*\$//')
+    END_VERSIONS
+    """
+}

modules/gatk4/genotypegvcfs/meta.yml

@@ -0,0 +1,69 @@
+name: gatk4_genotypegvcfs
+description: |
+  Perform joint genotyping on one or more samples pre-called with HaplotypeCaller.
+keywords:
+  - joint genotyping
+  - genotype
+  - gvcf
+tools:
+  - gatk4:
+      description: Genome Analysis Toolkit (GATK4)
+      homepage: https://gatk.broadinstitute.org/hc/en-us
+      documentation: https://gatk.broadinstitute.org/hc/en-us/categories/360002369672s
+      tool_dev_url: https://github.com/broadinstitute/gatk
+      doi: "10.1158/1538-7445.AM2017-3590"
+      licence: ['BSD-3-clause']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - gvcf:
+      type: tuple of files
+      description: |
+        Tuple of gVCF(.gz) file (first) and its index (second) or the path to a GenomicsDB (and empty)
+      pattern: ["*.{vcf,vcf.gz}", "*.{idx,tbi}"]
+  - fasta:
+      type: file
+      description: Reference fasta file
+      pattern: "*.fasta"
+  - fasta_index:
+      type: file
+      description: Reference fasta index file
+      pattern: "*.fai"
+  - fasta_dict:
+      type: file
+      description: Reference fasta sequence dict file
+      pattern: "*.dict"
+  - dbsnp:
+      type: file
+      description: dbSNP VCF file
+      pattern: "*.vcf.gz"
+  - dbsnp_index:
+      type: tuple of files
+      description: dbSNP VCF index file
+      pattern: "*.tbi"
+  - intervals_bed:
+      type: file
+      description: An intevals BED file
+      pattern: "*.bed"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - vcf:
+      type: file
+      description: Genotyped VCF file
+      pattern: "*.vcf.gz"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+
+authors:
+  - "@santiagorevale"

modules/gatk4/indexfeaturefile/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/gatk4/indexfeaturefile/main.nf

@@ -0,0 +1,40 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process GATK4_INDEXFEATUREFILE {
+    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::gatk4=4.2.0.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/gatk4:4.2.0.0--0"
+    } else {
+        container "quay.io/biocontainers/gatk4:4.2.0.0--0"
+    }
+
+    input:
+    tuple val(meta), path(feature_file)
+
+    output:
+    tuple val(meta), path("*.{tbi,idx}"), emit: index
+    path  "versions.yml"                , emit: versions
+
+    script:
+    """
+    gatk \\
+        IndexFeatureFile \\
+        $options.args \\
+        -I $feature_file
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$(echo \$(gatk --version 2>&1) | sed 's/^.*(GATK) v//; s/ .*\$//')
+    END_VERSIONS
+    """
+}

modules/gatk4/indexfeaturefile/meta.yml

@@ -0,0 +1,42 @@
+name: gatk4_indexfeaturefile
+description: Creates an index for a feature file, e.g. VCF or BED file.
+keywords:
+  - index
+  - feature
+tools:
+  - gatk4:
+      description: Genome Analysis Toolkit (GATK4)
+      homepage: https://gatk.broadinstitute.org/hc/en-us
+      documentation: https://gatk.broadinstitute.org/hc/en-us/categories/360002369672s
+      tool_dev_url: https://github.com/broadinstitute/gatk
+      doi: "10.1158/1538-7445.AM2017-3590"
+      licence: ['BSD-3-clause']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - feature_file:
+      type: file
+      description: VCF/BED file
+      pattern: "*.{vcf,vcf.gz,bed,bed.gz}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - index:
+      type: file
+      description: Index for VCF/BED file
+      pattern: "*.{tbi,idx}"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+
+authors:
+  - "@santiagorevale"

tests/config/pytest_modules.yml

@@ -474,6 +474,10 @@ gatk4/genomicsdbimport:
   - modules/gatk4/genomicsdbimport/**
   - tests/modules/gatk4/genomicsdbimport/**
 
+gatk4/genotypegvcfs:
+  - modules/gatk4/genotypegvcfs/**
+  - tests/modules/gatk4/genotypegvcfs/**
+
 gatk4/getpileupsummaries:
   - modules/gatk4/getpileupsummaries/**
   - tests/modules/gatk4/getpileupsummaries/**
@@ -482,6 +486,10 @@ gatk4/haplotypecaller:
   - modules/gatk4/haplotypecaller/**
   - tests/modules/gatk4/haplotypecaller/**
 
+gatk4/indexfeaturefile:
+  - modules/gatk4/indexfeaturefile/**
+  - tests/modules/gatk4/indexfeaturefile/**
+
 gatk4/intervallisttools:
   - modules/gatk4/intervallisttools/**
   - tests/modules/gatk4/intervallisttools/**

tests/config/test_data.config

@@ -191,6 +191,8 @@ params {
                 test2_pileups_table                            = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/test2.pileups.table"
                 test_genomicsdb_tar_gz                         = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/test_genomicsdb.tar.gz"
 
+                test_genomicsdb_tar_gz                         = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/test_genomicsdb.tar.gz"
+
                 test_test2_paired_mutect2_calls_vcf_gz         = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/paired_mutect2_calls/test_test2_paired_mutect2_calls.vcf.gz"
                 test_test2_paired_mutect2_calls_vcf_gz_tbi     = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/paired_mutect2_calls/test_test2_paired_mutect2_calls.vcf.gz.tbi"
                 test_test2_paired_mutect2_calls_vcf_gz_stats   = "${test_data_dir}/genomics/homo_sapiens/illumina/gatk/paired_mutect2_calls/test_test2_paired_mutect2_calls.vcf.gz.stats"

tests/modules/gatk4/genotypegvcfs/main.nf

@@ -0,0 +1,180 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GATK4_GENOTYPEGVCFS } from '../../../../modules/gatk4/genotypegvcfs/main.nf' addParams( options: [suffix:'.genotyped'] )
+include { UNTAR               } from '../../../../modules/untar/main.nf'               addParams( options: [:] )
+
+// Basic parameters with uncompressed VCF input
+workflow test_gatk4_genotypegvcfs_vcf_input {
+    
+    input = [ [ id:'test' ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf'], checkIfExists: true),
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_idx'], checkIfExists: true) ]
+
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, [], [], [] )
+}
+
+// Basic parameters with compressed VCF input
+workflow test_gatk4_genotypegvcfs_gz_input {
+    
+    input = [ [ id:'test' ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz'], checkIfExists: true),
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz_tbi'], checkIfExists: true) ]
+
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, [], [], [] )
+}
+
+// Basic parameters + optional dbSNP
+workflow test_gatk4_genotypegvcfs_gz_input_dbsnp {
+    
+    input = [ [ id:'test' ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz'], checkIfExists: true),
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz_tbi'], checkIfExists: true) ]
+
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    dbsnp        = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz'], checkIfExists: true)
+    dbsnpIndex   = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz_tbi'], checkIfExists: true)
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, dbsnp, dbsnpIndex, [] )
+}
+
+// Basic parameters + optional intervals
+workflow test_gatk4_genotypegvcfs_gz_input_intervals {
+    
+    input = [ [ id:'test' ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz'], checkIfExists: true),
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz_tbi'], checkIfExists: true) ]
+
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    intervalsBed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, [], [], intervalsBed )
+}
+
+// Basic parameters + optional dbSNP + optional intervals
+workflow test_gatk4_genotypegvcfs_gz_input_dbsnp_intervals {
+    
+    input = [ [ id:'test' ], // meta map
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz'], checkIfExists: true),
+              file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz_tbi'], checkIfExists: true) ]
+
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    dbsnp        = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz'], checkIfExists: true)
+    dbsnpIndex   = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz_tbi'], checkIfExists: true)
+
+    intervalsBed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, dbsnp, dbsnpIndex, intervalsBed )
+}
+
+// Basic parameters with GenomicsDB input
+workflow test_gatk4_genotypegvcfs_gendb_input {
+
+    fasta           = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex      = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict       = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    test_genomicsdb = file(params.test_data['homo_sapiens']['illumina']['test_genomicsdb_tar_gz'], checkIfExists: true)
+
+    UNTAR ( test_genomicsdb )
+
+    Channel.of(file("mock_gvcf_index.txt")).set{mock_gvcf_index}
+    Channel
+        .of([ id:'test' ])
+        .combine(UNTAR.out.untar)
+        .combine(mock_gvcf_index)
+        .set{ input }
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, [], [], [] )
+}
+
+// Basic parameters with GenomicsDB + optional dbSNP
+workflow test_gatk4_genotypegvcfs_gendb_input_dbsnp {
+    
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    dbsnp        = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz'], checkIfExists: true)
+    dbsnpIndex   = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz_tbi'], checkIfExists: true)
+
+    test_genomicsdb = file(params.test_data['homo_sapiens']['illumina']['test_genomicsdb_tar_gz'], checkIfExists: true)
+
+    UNTAR ( test_genomicsdb )
+
+    Channel.of(file("mock_gvcf_index.txt")).set{mock_gvcf_index}
+    Channel
+        .of([ id:'test' ])
+        .combine(UNTAR.out.untar)
+        .combine(mock_gvcf_index)
+        .set{ input }
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, dbsnp, dbsnpIndex, [] )
+}
+
+// Basic parameters with GenomicsDB + optional intervals
+workflow test_gatk4_genotypegvcfs_gendb_input_intervals {
+    
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    intervalsBed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+
+    test_genomicsdb = file(params.test_data['homo_sapiens']['illumina']['test_genomicsdb_tar_gz'], checkIfExists: true)
+
+    UNTAR ( test_genomicsdb )
+
+    Channel.of(file("mock_gvcf_index.txt")).set{mock_gvcf_index}
+    Channel
+        .of([ id:'test' ])
+        .combine(UNTAR.out.untar)
+        .combine(mock_gvcf_index)
+        .set{ input }
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, [], [], intervalsBed )
+}
+
+// Basic parameters with GenomicsDB + optional dbSNP + optional intervals
+workflow test_gatk4_genotypegvcfs_gendb_input_dbsnp_intervals {
+    
+    fasta        = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fastaIndex   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    fastaDict    = file(params.test_data['homo_sapiens']['genome']['genome_dict'], checkIfExists: true)
+
+    dbsnp        = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz'], checkIfExists: true)
+    dbsnpIndex   = file(params.test_data['homo_sapiens']['genome']['dbsnp_146_hg38_vcf_gz_tbi'], checkIfExists: true)
+
+    intervalsBed = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+
+    test_genomicsdb = file(params.test_data['homo_sapiens']['illumina']['test_genomicsdb_tar_gz'], checkIfExists: true)
+
+    UNTAR ( test_genomicsdb )
+
+    Channel.of(file("mock_gvcf_index.txt")).set{mock_gvcf_index}
+    Channel
+        .of([ id:'test' ])
+        .combine(UNTAR.out.untar)
+        .combine(mock_gvcf_index)
+        .set{ input }
+
+    GATK4_GENOTYPEGVCFS ( input, fasta, fastaIndex, fastaDict, dbsnp, dbsnpIndex, intervalsBed )
+}

tests/modules/gatk4/genotypegvcfs/test.yml

@@ -0,0 +1,80 @@
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_vcf_input
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_vcf_input -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gz_input
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gz_input -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gz_input_dbsnp
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gz_input_dbsnp -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DB;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gz_input_intervals
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gz_input_intervals -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gz_input_dbsnp_intervals
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gz_input_dbsnp_intervals -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DB;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gendb_input
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gendb_input -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gendb_input_dbsnp
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gendb_input_dbsnp -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DB;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gendb_input_intervals
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gendb_input_intervals -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']
+
+- name: gatk4 genotypegvcfs test_gatk4_genotypegvcfs_gendb_input_dbsnp_intervals
+  command: nextflow run tests/modules/gatk4/genotypegvcfs -entry test_gatk4_genotypegvcfs_gendb_input_dbsnp_intervals -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/genotypegvcfs
+  files:
+    - path: output/gatk4/test.genotyped.vcf.gz
+      contains: ['AC=1;AF=0.500;AN=2;BaseQRankSum=0.00;DB;DP=211;ExcessHet=3.0103;FS=0.000;MLEAC=1;MLEAF=0.500;MQ=60.00;MQRankSum=0.00;QD=0.95;ReadPosRankSum=1.09;SOR=0.680']

tests/modules/gatk4/indexfeaturefile/main.nf

@@ -0,0 +1,45 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GATK4_INDEXFEATUREFILE } from '../../../../modules/gatk4/indexfeaturefile/main.nf' addParams( options: [:] )
+
+workflow test_gatk4_indexfeaturefile_bed {
+    
+    input = [
+        [ id:'test' ], // meta map
+        file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true) 
+    ]
+
+    GATK4_INDEXFEATUREFILE ( input )
+}
+
+workflow test_gatk4_indexfeaturefile_bed_gz {
+    
+    input = [
+        [ id:'test' ], // meta map
+        file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true)
+    ]
+
+    GATK4_INDEXFEATUREFILE ( input )
+}
+
+workflow test_gatk4_indexfeaturefile_vcf {
+    
+    input = [ 
+        [ id:'test' ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf'], checkIfExists: true) 
+    ]
+
+    GATK4_INDEXFEATUREFILE ( input )
+}
+
+workflow test_gatk4_indexfeaturefile_vcf_gz {
+    
+    input = [ 
+        [ id:'test' ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_genome_vcf_gz'], checkIfExists: true) 
+    ]
+
+    GATK4_INDEXFEATUREFILE ( input )
+}

tests/modules/gatk4/indexfeaturefile/test.yml

@@ -0,0 +1,39 @@
+# We can't use an md5sum or check file contents because:
+# a) the path to the file is embedded inside it, 
+# b) the file is binary so we can't check for text inside it.
+- name: gatk4 indexfeaturefile test_gatk4_indexfeaturefile_bed
+  command: nextflow run tests/modules/gatk4/indexfeaturefile -entry test_gatk4_indexfeaturefile_bed -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/indexfeaturefile
+  files:
+    - path: output/gatk4/genome.bed.idx
+
+- name: gatk4 indexfeaturefile test_gatk4_indexfeaturefile_bed_gz
+  command: nextflow run tests/modules/gatk4/indexfeaturefile -entry test_gatk4_indexfeaturefile_bed_gz -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/indexfeaturefile
+  files:
+    - path: output/gatk4/genome.bed.gz.tbi
+      md5sum: 2eb6ed0a0b049efe4caa1413089dcd74
+
+# We can't use an md5sum or check file contents because:
+# a) the path to the file is embedded inside it, 
+# b) the file is binary so we can't check for text inside it.
+- name: gatk4 indexfeaturefile test_gatk4_indexfeaturefile_vcf
+  command: nextflow run tests/modules/gatk4/indexfeaturefile -entry test_gatk4_indexfeaturefile_vcf -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/indexfeaturefile
+  files:
+    - path: output/gatk4/test.genome.vcf.idx
+
+- name: gatk4 indexfeaturefile test_gatk4_indexfeaturefile_vcf_gz
+  command: nextflow run tests/modules/gatk4/indexfeaturefile -entry test_gatk4_indexfeaturefile_vcf_gz -c tests/config/nextflow.config
+  tags:
+    - gatk4
+    - gatk4/indexfeaturefile
+  files:
+    - path: output/gatk4/test.genome.vcf.gz.tbi
+      md5sum: ea03cd1d1f178eefa656787537053c37