add freebayes/somatic and update freebayes/germline (#990)

* add freebayes/somatic and update freebayes/germline to new syntax and cram

* add pytest

* update with new freebayes/germline path

modules/freebayes/germline/main.nf

@@ -4,7 +4,7 @@ include { initOptions; saveFiles; getProcessName; getSoftwareName } from './func
 params.options = [:]
 options        = initOptions(params.options)
 
-process FREEBAYES {
+process FREEBAYES_GERMLINE {
     tag "$meta.id"
     label 'process_low'
     publishDir "${params.outdir}",
@@ -19,24 +19,25 @@ process FREEBAYES {
     }
 
     input:
-    tuple val(meta), path(bam), path(bai)
-    tuple path(fasta), path(fai)
-    path(targets)
-    path(samples)
-    path(populations)
-    path(cnv)
-
+    tuple val(meta), path(input), path(input_index)
+    path fasta
+    path fai
+    path targets
+    path samples
+    path populations
+    path cnv
 
     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 targets_file = targets ? "--target ${targets}" : ""
-    def samples_file = samples ? "--samples ${samples}" : ""
-    def populations_file = populations ? "--populations ${populations}" : ""
-    def cnv_file = cnv ? "--cnv-map ${cnv}" : ""
+    def prefix           = options.suffix ? "${meta.id}${options.suffix}"  : "${meta.id}"
+    def targets_file     = targets        ? "--target ${targets}"          : ""
+    def samples_file     = samples        ? "--samples ${samples}"         : ""
+    def populations_file = populations    ? "--populations ${populations}" : ""
+    def cnv_file         = cnv            ? "--cnv-map ${cnv}"             : ""
+
     if (task.cpus > 1) {
         """
         freebayes-parallel \\
@@ -47,7 +48,7 @@ process FREEBAYES {
             $populations_file \\
             $cnv_file \\
             $options.args \\
-            $bam  > ${prefix}.vcf
+            $input  > ${prefix}.vcf
 
         gzip --no-name ${prefix}.vcf
 
@@ -66,7 +67,7 @@ process FREEBAYES {
             $populations_file \\
             $cnv_file \\
             $options.args \\
-            $bam > ${prefix}.vcf
+            $input > ${prefix}.vcf
 
         gzip --no-name ${prefix}.vcf
 

modules/freebayes/germline/meta.yml

@@ -1,4 +1,4 @@
-name: freebayes
+name: freebayes_germline
 description: A haplotype-based variant detector
 keywords:
   - variant caller
@@ -12,7 +12,7 @@ tools:
       homepage: https://github.com/freebayes/freebayes
       documentation: https://github.com/freebayes/freebayes
       tool_dev_url: https://github.com/freebayes/freebayes
-      doi: ""
+      doi: "arXiv:1207.3907"
       licence: ['MIT']
 
 input:
@@ -21,11 +21,11 @@ input:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - bam:
+  - input:
       type: file
       description: BAM/CRAM/SAM file
       pattern: "*.{bam,cram,sam}"
-  - bai:
+  - input_index:
       type: file
       description: BAM/CRAM/SAM index file
       pattern: "*.bam.bai"
@@ -59,7 +59,6 @@ input:
           seq_name start end sample_name copy_number
       pattern: "*.bed"
 
-
 output:
   - meta:
       type: map
@@ -76,3 +75,4 @@ output:
       pattern: "*.vcf.gz"
 authors:
   - "@maxibor"
+  - "@FriederikeHanssen"

modules/freebayes/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/freebayes/somatic/main.nf

@@ -0,0 +1,74 @@
+// Import generic module functions
+include { initOptions; saveFiles; getProcessName; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process FREEBAYES_SOMATIC {
+    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::freebayes=1.3.5" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/freebayes:1.3.5--py38ha193a2f_3"
+    } else {
+        container "quay.io/biocontainers/freebayes:1.3.5--py38ha193a2f_3"
+    }
+
+    input:
+    tuple val(meta), path(input_normal), path(input_index_normal), path(input_tumor), path(input_index_tumor)
+    path fasta
+    path fai
+    path targets
+    path samples
+
+    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 targets_file     = targets        ? "--target ${targets}"          : ""
+    def samples_file     = samples        ? "--samples ${samples}"         : ""
+
+    if (task.cpus > 1) {
+        """
+        freebayes-parallel \\
+            <(fasta_generate_regions.py ${fasta}.fai 10000) ${task.cpus} \\
+            -f $fasta \\
+            $targets_file \\
+            $samples_file \\
+            $options.args \\
+            $input_tumor \\
+            $input_normal  > ${prefix}.vcf
+
+        gzip --no-name ${prefix}.vcf
+
+        cat <<-END_VERSIONS > versions.yml
+        ${getProcessName(task.process)}:
+            ${getSoftwareName(task.process)}: \$(echo \$(freebayes --version 2>&1) | sed 's/version:\s*v//g' )
+        END_VERSIONS
+        """
+
+    } else {
+        """
+        freebayes \\
+            -f $fasta \\
+            $targets_file \\
+            $samples_file \\
+            $options.args \\
+            $input_tumor \\
+            $input_normal  > ${prefix}.vcf
+
+        gzip --no-name ${prefix}.vcf
+
+        cat <<-END_VERSIONS > versions.yml
+        ${getProcessName(task.process)}:
+            ${getSoftwareName(task.process)}: \$(echo \$(freebayes --version 2>&1) | sed 's/version:\s*v//g' )
+        END_VERSIONS
+        """
+    }
+}

modules/freebayes/somatic/meta.yml

@@ -0,0 +1,66 @@
+name: freebayes_somatic
+description: A haplotype-based variant detector
+keywords:
+  - variant caller
+  - SNP
+  - genotyping
+  - somatic variant calling
+  - bayesian
+
+tools:
+  - freebayes:
+      description: Bayesian haplotype-based polymorphism discovery and genotyping
+      homepage: https://github.com/freebayes/freebayes
+      documentation: https://github.com/freebayes/freebayes
+      tool_dev_url: https://github.com/freebayes/freebayes
+      doi: "arXiv:1207.3907"
+      licence: ['MIT']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input:
+      type: file
+      description: BAM/CRAM/SAM file
+      pattern: "*.{bam,cram,sam}"
+  - input_index:
+      type: file
+      description: BAM/CRAM/SAM index file
+      pattern: "*.bam.bai"
+  - fasta:
+      type: file
+      description: reference fasta file
+      pattern: ".{fa,fa.gz,fasta,fasta.gz}"
+  - fai:
+      type: file
+      description: reference fasta file index
+      pattern: "*.fai"
+  - targets:
+      type: file
+      description: Optional - Limit analysis to targets listed in this BED-format FILE.
+      pattern: "*.bed"
+  - samples:
+      type: file
+      description: Optional - Limit analysis to samples listed (one per line) in the FILE.
+      pattern: "*.txt"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "*.{version.txt}"
+  - vcf:
+      type: file
+      description: Compressed VCF file
+      pattern: "*.vcf.gz"
+
+authors:
+  - "@FriederikeHanssen"

modules/gatk4/applybqsr/meta.yml

@@ -61,3 +61,4 @@ output:
 
 authors:
   - "@yocra3"
+  - "@FriederikeHanssen"

tests/config/pytest_modules.yml

@@ -410,9 +410,13 @@ flash:
   - modules/flash/**
   - tests/modules/flash/**
 
-freebayes:
-  - modules/freebayes/**
-  - tests/modules/freebayes/**
+freebayes/germline:
+  - modules/freebayes/germline/**
+  - tests/modules/freebayes/germline/**
+
+freebayes/somatic:
+  - modules/freebayes/somatic/**
+  - tests/modules/freebayes/somatic/**
 
 gatk4/applybqsr:
   - modules/gatk4/applybqsr/**

tests/modules/freebayes/germline/main.nf

@@ -0,0 +1,51 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { FREEBAYES_GERMLINE } from '../../../../modules/freebayes/germline/main.nf' addParams( options: [:] )
+
+workflow test_freebayes {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
+    fasta   = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fai         = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    targets = []
+    samples = []
+    populations = []
+    cnv = []
+
+    FREEBAYES_GERMLINE ( input, fasta, fai, targets, samples, populations, cnv)
+}
+
+workflow test_freebayes_bed {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
+    fasta   = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fai         = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+    targets     = file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)
+    samples     = []
+    populations = []
+    cnv         = []
+
+    FREEBAYES_GERMLINE ( input, fasta, fai, targets, samples, populations, cnv)
+}
+
+workflow test_freebayes_cram {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_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)
+    targets = []
+    samples = []
+    populations = []
+    cnv = []
+
+    FREEBAYES_GERMLINE ( input, fasta, fai, targets, samples, populations, cnv)
+}

tests/modules/freebayes/germline/test.yml

@@ -0,0 +1,26 @@
+- name: freebayes germline test_freebayes
+  command: nextflow run tests/modules/freebayes/germline -entry test_freebayes -c tests/config/nextflow.config
+  tags:
+    - freebayes
+    - freebayes/germline
+  files:
+    - path: output/freebayes/test.vcf.gz
+      md5sum: 1ec210ad27514c7a4140c924dc66d979
+
+- name: freebayes germline test_freebayes_bed
+  command: nextflow run tests/modules/freebayes/germline -entry test_freebayes_bed -c tests/config/nextflow.config
+  tags:
+    - freebayes
+    - freebayes/germline
+  files:
+    - path: output/freebayes/test.vcf.gz
+      md5sum: e8923cccd5dac196f72d3d3997a60706
+
+- name: freebayes germline test_freebayes_cram
+  command: nextflow run tests/modules/freebayes/germline -entry test_freebayes_cram -c tests/config/nextflow.config
+  tags:
+    - freebayes
+    - freebayes/germline
+  files:
+    - path: output/freebayes/test.vcf.gz
+      md5sum: cb57a3ed154618e3aa4a5272fcfb7521

tests/modules/freebayes/main.nf

@@ -1,35 +0,0 @@
-#!/usr/bin/env nextflow
-
-nextflow.enable.dsl = 2
-
-include { FREEBAYES } from '../../../modules/freebayes/main.nf' addParams( options: [:] )
-
-workflow test_freebayes {
-
-    input = [ [ id:'test', single_end:false ], // meta map
-              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
-              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
-    reference = [file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true),
-                 file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)]
-    targets = []
-    samples = []
-    populations = []
-    cnv = []
-
-    FREEBAYES ( input, reference, targets, samples, populations, cnv)
-}
-
-workflow test_freebayes_bed {
-
-    input = [ [ id:'test', single_end:false ], // meta map
-              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
-              file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
-    reference = [file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true),
-                 file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)]
-    targets = file(params.test_data['sarscov2']['genome']['test_bed'], checkIfExists: true)
-    samples = []
-    populations = []
-    cnv = []
-
-    FREEBAYES ( input, reference, targets, samples, populations, cnv)
-}

tests/modules/freebayes/somatic/main.nf

@@ -0,0 +1,37 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { FREEBAYES_SOMATIC } from '../../../../modules/freebayes/somatic/main.nf' addParams( options: [:] )
+
+workflow test_freebayes {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_bam_bai'], 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)
+    targets = []
+    samples = []
+
+    FREEBAYES_SOMATIC ( input, fasta, fai, targets, samples)
+}
+
+workflow test_freebayes_intervals {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_sorted_cram_crai'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_sorted_cram'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test2_paired_end_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)
+    targets = file(params.test_data['homo_sapiens']['genome']['genome_bed'], checkIfExists: true)
+    samples = []
+
+    FREEBAYES_SOMATIC ( input, fasta, fai, targets, samples)
+}

tests/modules/freebayes/somatic/test.yml

@@ -0,0 +1,17 @@
+- name: freebayes somatic test_freebayes
+  command: nextflow run tests/modules/freebayes/somatic -entry test_freebayes -c tests/config/nextflow.config
+  tags:
+    - freebayes/somatic
+    - freebayes
+  files:
+    - path: output/freebayes/test.vcf.gz
+      md5sum: 1c47d02f27ec5918558c8688ce6e7780
+
+- name: freebayes somatic test_freebayes_intervals
+  command: nextflow run tests/modules/freebayes/somatic -entry test_freebayes_intervals -c tests/config/nextflow.config
+  tags:
+    - freebayes/somatic
+    - freebayes
+  files:
+    - path: output/freebayes/test.vcf.gz
+      md5sum: 5b8a12666bde63746dcec7afcd3ef789

tests/modules/freebayes/test.yml

@@ -1,14 +0,0 @@
-- name: freebayes test_freebayes
-  command: nextflow run tests/modules/freebayes -entry test_freebayes -c tests/config/nextflow.config
-  tags:
-    - freebayes
-  files:
-    - path: output/freebayes/test.vcf.gz
-
-- name: freebayes test_freebayes_bed
-  command: nextflow run tests/modules/freebayes -entry test_freebayes_bed -c tests/config/nextflow.config
-  tags:
-    - freebayes
-  files:
-    - path: output/freebayes/test.vcf.gz
-