add genmap/mappability (#509)

* add genmap/mappability

* Add index module.
Remove -w parameter from mappability module.

* change the output of genmap/index

* fix the lint error which can not handle stageAs.

* Apply suggestions from code review

* Update main.nf

* Update test.yml

* Update main.nf

* Update test.yml

* Update tests/software/genmap/mappability/main.nf

* Update software/genmap/mappability/main.nf

* Update software/genmap/mappability/main.nf

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

software/genmap/index/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"
+        }
+    }
+}

software/genmap/index/main.nf

@@ -0,0 +1,38 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process GENMAP_INDEX {
+    tag '$fasta'
+    label 'process_high'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:'index', meta:[:], publish_by_meta:[]) }
+
+    conda (params.enable_conda ? "bioconda::genmap=1.3.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/genmap:1.3.0--h1b792b2_1"
+    } else {
+        container "quay.io/biocontainers/genmap:1.3.0--h1b792b2_1"
+    }
+
+    input:
+    path fasta
+
+    output:
+    path "genmap"       , emit: index
+    path "*.version.txt", emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    """
+    genmap \\
+        index \\
+        -F $fasta \\
+        -I genmap
+
+    echo \$(genmap --version 2>&1) | sed 's/GenMap version: //; s/SeqAn.*\$//' > ${software}.version.txt
+    """
+}

software/genmap/index/meta.yml

@@ -0,0 +1,31 @@
+name: genmap_index
+description: create index file for genmap
+keywords:
+  - index
+tools:
+  - genmap:
+      description: Ultra-fast computation of genome mappability.
+      homepage: https://github.com/cpockrandt/genmap
+      documentation: https://github.com/cpockrandt/genmap
+      tool_dev_url: https://github.com/cpockrandt/genmap
+      doi: "10.1093/bioinformatics/btaa222"
+      licence: ['BSD']
+
+input:
+  - fasta:
+      type: file
+      description: fasta file
+      pattern: "*.{fasta,fa}"
+
+output:
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "*.{version.txt}"
+  - index:
+      type: index
+      description: Genmap index file
+      pattern: "{index}.*"
+
+authors:
+  - "@jianhong"

software/genmap/mappability/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"
+        }
+    }
+}

software/genmap/mappability/main.nf

@@ -0,0 +1,41 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process GENMAP_MAPPABILITY {
+    tag '$fasta'
+    label 'process_high'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:[:], publish_by_meta:[]) }
+
+    conda (params.enable_conda ? "bioconda::genmap=1.3.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/genmap:1.3.0--h1b792b2_1"
+    } else {
+        container "quay.io/biocontainers/genmap:1.3.0--h1b792b2_1"
+    }
+
+    input:
+    path index
+
+    output:
+    path "*.wig"        , optional:true, emit: wig
+    path "*.bedgraph"   , optional:true, emit: bedgraph
+    path "*.txt"        , optional:true, emit: txt
+    path "*.version.txt"               , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    """
+    genmap \\
+        map \\
+        $options.args \\
+        -I $index \\
+        -O mappability
+
+    echo \$(genmap --version 2>&1) | sed 's/GenMap version: //; s/SeqAn.*\$//' > ${software}.version.txt
+    """
+}

software/genmap/mappability/meta.yml

@@ -0,0 +1,42 @@
+name: genmap_mappability
+description: create mappability file for a genome
+keywords:
+  - mappability
+tools:
+  - genmap:
+      description: Ultra-fast computation of genome mappability.
+      homepage: https://github.com/cpockrandt/genmap
+      documentation: https://github.com/cpockrandt/genmap
+      tool_dev_url: https://github.com/cpockrandt/genmap
+      doi: "10.1093/bioinformatics/btaa222"
+      licence: ['BSD']
+
+input:
+  - fasta:
+      type: file
+      description: fasta file
+      pattern: "*.{fasta,fa}"
+  - index:
+      type: file
+      description: index file
+
+output:
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "*.{version.txt}"
+  - wig:
+      type: file
+      description: genmap wig mappability file
+      pattern: "*.wig"
+  - bedgraph:
+      type: file
+      description: genmap bedgraph mappability file
+      pattern: "*.bedgraph"
+  - txt:
+      type: file
+      description: genmap text mappability file
+      pattern: "*.txt"
+
+authors:
+  - "@jianhong"

tests/config/pytest_software.yml

@@ -294,6 +294,14 @@ gatk4/variantfiltration:
   - software/gatk4/variantfiltration/**
   - tests/software/gatk4/variantfiltration/**
 
+genmap/index:
+  - software/genmap/index/**
+  - tests/software/genmap/index/**
+
+genmap/mappability:
+  - software/genmap/mappability/**
+  - tests/software/genmap/mappability/**
+
 gffread:
   - software/gffread/**
   - tests/software/gffread/**

tests/software/genmap/index/main.nf

@@ -0,0 +1,12 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMAP_INDEX } from '../../../../software/genmap/index/main.nf' addParams( options: [:] )
+
+workflow test_genmap_index {
+
+    input = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    GENMAP_INDEX ( input )
+}

tests/software/genmap/index/test.yml

@@ -0,0 +1,48 @@
+- name: genmap index test_genmap_index
+  command: nextflow run tests/software/genmap/index -entry test_genmap_index -c tests/config/nextflow.config
+  tags:
+    - genmap
+    - genmap/index
+  files:
+    - path: output/index/genmap/index.ids.concat
+      md5sum: da6caa25f62c5407ccdfbcce1fa92408
+    - path: output/index/genmap/index.ids.limits
+      md5sum: f82636c5da188aec131d3a809473eff1
+    - path: output/index/genmap/index.info.concat
+      md5sum: 8ba5273aa9e58722bf45b9cc39fc6bfe
+    - path: output/index/genmap/index.info.limits
+      md5sum: 3522f2811f4ddf04598809fc84a1459e
+    - path: output/index/genmap/index.lf.drp
+      md5sum: dd85d6a23af2c7adf2695658e3056c08
+    - path: output/index/genmap/index.lf.drp.sbl
+      md5sum: f1d3ff8443297732862df21dc4e57262
+    - path: output/index/genmap/index.lf.drs
+      md5sum: 93b885adfe0da089cdf634904fd59f71
+    - path: output/index/genmap/index.lf.drv
+      md5sum: e06b605496bd91b32afa3c4f56d934ac
+    - path: output/index/genmap/index.lf.drv.sbl
+      md5sum: 8dd6bb7329a71449b0a1b292b5999164
+    - path: output/index/genmap/index.lf.pst
+      md5sum: e8daba34298e99e42942435286f9b3f0
+    - path: output/index/genmap/index.rev.lf.drp
+      md5sum: 5d9107e3aeec0721553dd661d4365fef
+    - path: output/index/genmap/index.rev.lf.drp.sbl
+      md5sum: f1d3ff8443297732862df21dc4e57262
+    - path: output/index/genmap/index.rev.lf.drs
+      md5sum: 93b885adfe0da089cdf634904fd59f71
+    - path: output/index/genmap/index.rev.lf.drv
+      md5sum: df7e795edc0a034577a9d2599fe8cfeb
+    - path: output/index/genmap/index.rev.lf.drv.sbl
+      md5sum: 8dd6bb7329a71449b0a1b292b5999164
+    - path: output/index/genmap/index.rev.lf.pst
+      md5sum: e8daba34298e99e42942435286f9b3f0
+    - path: output/index/genmap/index.sa.ind
+      md5sum: e21e5c7ce887cc8e3d0fa44ab1019cab
+    - path: output/index/genmap/index.sa.len
+      md5sum: 5dfc20cfe8ed9892451461a8d402f51c
+    - path: output/index/genmap/index.sa.val
+      md5sum: 400ee7f2fe93b2000ae3a5da5e509730
+    - path: output/index/genmap/index.txt.concat
+      md5sum: b4303962e0c176107945f3405370e6ae
+    - path: output/index/genmap/index.txt.limits
+      md5sum: 4480a068db603e4c9a27bc4fa9ceaf14

tests/software/genmap/mappability/main.nf

@@ -0,0 +1,14 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GENMAP_INDEX       } from '../../../../software/genmap/index/main.nf'       addParams( options: [:] )
+include { GENMAP_MAPPABILITY } from '../../../../software/genmap/mappability/main.nf' addParams( options: [args : '-K 50 -E 2 -w -t -bg'] )
+
+workflow test_genmap_map {
+
+    input = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+
+    GENMAP_INDEX ( input )
+    GENMAP_MAPPABILITY ( GENMAP_INDEX.out.index )
+}

tests/software/genmap/mappability/test.yml

@@ -0,0 +1,12 @@
+- name: genmap mappability test_genmap_map
+  command: nextflow run tests/software/genmap/mappability -entry test_genmap_map -c tests/config/nextflow.config
+  tags:
+    - genmap
+    - genmap/mappability
+  files:
+    - path: output/genmap/mappability.bedgraph
+      md5sum: c4203b84a84331282ef8660279f80b26
+    - path: output/genmap/mappability.txt
+      md5sum: 602d409493457f92b2e5f075f62ebe30
+    - path: output/genmap/mappability.wig
+      md5sum: fd2afb6512ab9e86f9a3b62f8b9d9088