Add strelka/somatic module (#866)

* Add strelka/somatic module

* Fill out meta.yml properly

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+    // Do not publish versions.yml unless running from pytest workflow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

@@ -0,0 +1,59 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process STRELKA_SOMATIC {
+    tag "$meta.id"
+    label 'process_high'
+    publishDir "${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::strelka=2.9.10" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/strelka:2.9.10--h9ee0642_1"
+    } else {
+        container "quay.io/biocontainers/strelka:2.9.10--h9ee0642_1"
+    }
+
+    input:
+    tuple val(meta), path(cram_normal), path(crai_normal), path(cram_tumor), path(crai_tumor)
+    path  fasta
+    path  fai
+    path  target_bed
+    path  target_bed_tbi
+
+    output:
+    tuple val(meta), path("*.somatic_indels.vcf.gz")    , emit: vcf_indels
+    tuple val(meta), path("*.somatic_indels.vcf.gz.tbi"), emit: vcf_indels_tbi
+    tuple val(meta), path("*.somatic_snvs.vcf.gz")      , emit: vcf_snvs
+    tuple val(meta), path("*.somatic_snvs.vcf.gz.tbi")  , emit: vcf_snvs_tbi
+    path "versions.yml"                                 , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def options_strelka = params.target_bed ? "--exome --callRegions ${target_bed}" : ""
+    """
+    configureStrelkaSomaticWorkflow.py \\
+        --tumor $cram_tumor \\
+        --normal $cram_normal \\
+        --referenceFasta $fasta \\
+        $options_strelka \\
+        $options.args \\
+        --runDir strelka
+
+    python strelka/runWorkflow.py -m local -j $task.cpus
+
+    mv strelka/results/variants/somatic.indels.vcf.gz     ${prefix}.somatic_indels.vcf.gz
+    mv strelka/results/variants/somatic.indels.vcf.gz.tbi ${prefix}.somatic_indels.vcf.gz.tbi
+    mv strelka/results/variants/somatic.snvs.vcf.gz       ${prefix}.somatic_snvs.vcf.gz
+    mv strelka/results/variants/somatic.snvs.vcf.gz.tbi   ${prefix}.somatic_snvs.vcf.gz.tbi
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$( configureStrelkaSomaticWorkflow.py --version )
+    END_VERSIONS
+    """
+}

modules/strelka/somatic/meta.yml

@@ -0,0 +1,85 @@
+name: strelka_somatic
+description: Strelka2 is a fast and accurate small variant caller optimized for analysis of germline variation in small cohorts and somatic variation in tumor/normal sample pairs
+keywords:
+  - variant calling
+  - germline
+  - wgs
+  - vcf
+  - variants
+tools:
+  - strelka:
+      description: Strelka calls somatic and germline small variants from mapped sequencing reads
+      homepage: https://github.com/Illumina/strelka
+      documentation: https://github.com/Illumina/strelka/blob/v2.9.x/docs/userGuide/README.md
+      tool_dev_url: https://github.com/Illumina/strelka
+      doi: 10.1038/s41592-018-0051-x
+      licence: ['GPL v3']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - cram_normal:
+      type: file
+      description: BAM/CRAM/SAM file
+      pattern: "*.{bam,cram,sam}"
+  - crai_normal:
+      type: file
+      description: BAM/CRAM/SAM index file
+      pattern: "*.{bai,crai,sai}"
+  - cram_tumor:
+      type: file
+      description: BAM/CRAM/SAM file
+      pattern: "*.{bam,cram,sam}"
+  - crai_tumor:
+      type: file
+      description: BAM/CRAM/SAM index file
+      pattern: "*.{bai,crai,sai}"
+  - fasta:
+      type: file
+      description: Genome reference FASTA file
+      pattern: "*.{fa,fasta}"
+  - fai:
+      type: file
+      description: Genome reference FASTA index file
+      pattern: "*.{fa.fai,fasta.fai}"
+  - target_bed:
+      type: file
+      description: BED file containing target regions for variant calling
+      pattern: "*.{bed}"
+  - target_bed_tbi:
+      type: file
+      description: Index for BED file containing target regions for variant calling
+      pattern: "*.{bed.tbi}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - vcf_indels:
+      type: file
+      description: Gzipped VCF file containing variants
+      pattern: "*.{vcf.gz}"
+  - vcf_indels_tbi:
+      type: file
+      description: Index for gzipped VCF file containing variants
+      pattern: "*.{vcf.gz.tbi}"
+  - vcf_snvs:
+      type: file
+      description: Gzipped VCF file containing variants
+      pattern: "*.{vcf.gz}"
+  - vcf_snvs_tbi:
+      type: file
+      description: Index for gzipped VCF file containing variants
+      pattern: "*.{vcf.gz.tbi}"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+
+authors:
+  - "@drpatelh"

tests/config/pytest_modules.yml

@@ -991,14 +991,14 @@ spatyper:
   - modules/spatyper/**
   - tests/modules/spatyper/**
 
-sratools/prefetch:
-  - modules/sratools/prefetch/**
-  - tests/modules/sratools/prefetch/**
-
 sratools/fasterqdump:
   - modules/sratools/fasterqdump/**
   - tests/modules/sratools/fasterqdump/**
 
+sratools/prefetch:
+  - modules/sratools/prefetch/**
+  - tests/modules/sratools/prefetch/**
+
 staphopiasccmec:
   - modules/staphopiasccmec/**
   - tests/modules/staphopiasccmec/**
@@ -1015,6 +1015,10 @@ strelka/germline:
   - modules/strelka/germline/**
   - tests/modules/strelka/germline/**
 
+strelka/somatic:
+  - modules/strelka/somatic/**
+  - tests/modules/strelka/somatic/**
+
 stringtie/merge:
   - modules/stringtie/merge/**
   - tests/modules/stringtie/merge/**

tests/modules/strelka/somatic/main.nf

@@ -0,0 +1,23 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { STRELKA_SOMATIC } from '../../../../modules/strelka/somatic/main.nf' addParams( options: [:] )
+
+workflow test_strelka_somatic {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
+        file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true)
+    ]
+
+    fasta   = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fai     = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+    bed     = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz'], checkIfExists: true)
+    bed_tbi = file(params.test_data['homo_sapiens']['genome']['genome_bed_gz_tbi'], checkIfExists: true)
+
+    STRELKA_SOMATIC ( input, fasta, fai, bed, bed_tbi )
+}

tests/modules/strelka/somatic/test.yml

@@ -0,0 +1,12 @@
+- name: strelka somatic test_strelka_somatic
+  command: nextflow run tests/modules/strelka/somatic -entry test_strelka_somatic -c tests/config/nextflow.config
+  tags:
+    - strelka
+    - strelka/somatic
+  files:
+    - path: output/strelka/test.somatic_indels.vcf.gz
+    - path: output/strelka/test.somatic_indels.vcf.gz.tbi
+      md5sum: 4cb176febbc8c26d717a6c6e67b9c905
+    - path: output/strelka/test.somatic_snvs.vcf.gz
+    - path: output/strelka/test.somatic_snvs.vcf.gz.tbi
+      md5sum: 4cb176febbc8c26d717a6c6e67b9c905