Add flye module

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+    // Do not publish versions.yml unless running from pytest workflow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

@@ -0,0 +1,56 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process FLYE {
+    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 ? "flye==2.9--py38h69e0bdc_0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/flye:2.9--py39h39abbe0_0"
+    } else {
+        container "quay.io/biocontainers/flye:2.9--py38h69e0bdc_0"
+    }
+
+    input:
+    tuple val(meta), path(reads)
+
+    output:
+    tuple val(meta), path("*.fasta"), emit: fasta
+    tuple val(meta), path("*.gfa")  , emit: gfa
+    tuple val(meta), path("*.gv")   , emit: gv
+    tuple val(meta), path("*.txt")  , emit: txt
+    tuple val(meta), path("*.log")  , emit: log
+    tuple val(meta), path("*.json") , emit: json
+    path "versions.yml"             , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def mode = meta.mode
+    """
+    flye \\
+        --$mode \\
+        $reads \\
+        $options.args \\
+        --threads $task.cpus \\
+        --out-dir ./
+
+    mv assembly.fasta ${prefix}.assembly.fasta
+    mv assembly_graph.gfa ${prefix}.assembly_graph.gfa
+    mv assembly_graph.gv ${prefix}.assembly_graph.gv
+    mv assembly_info.txt ${prefix}.assembly_info.txt
+    mv flye.log ${prefix}.flye.log
+    mv params.json ${prefix}.params.json
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$( flye --version | sed 's/-b1768//' )
+    END_VERSIONS
+    """
+}

modules/flye/meta.yml

@@ -0,0 +1,65 @@
+name: flye
+description: De novo assembler for single molecule sequencing reads
+keywords:
+  - assembly
+  - genome
+  - de novo
+  - genome assembler
+  - single molecule
+tools:
+  - flye:
+      description: De novo assembler for single molecule sequencing reads using repeat graphs
+      homepage: https://github.com/fenderglass/Flye
+      documentation: https://github.com/fenderglass/Flye/blob/flye/docs/USAGE.md
+      tool_dev_url: https://github.com/fenderglass/Flye
+      doi: doi:s41592-020-00971-x
+      licence: ['BSD-3-clause']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', mode:'pacbio-raw' ]
+  - reads:
+      type: file
+      description: Input file in FASTA/FASTQ format.
+      pattern: "*.{fasta,fastq,fasta.gz,fastq.gz,fa,fq,fa.gz,fq.gz}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', mode:pacbio-raw ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - fasta:
+      type: file
+      description: Assembled FASTA file
+      pattern: "*.fasta"
+  - gfa:
+      type: file
+      description: Repeat graph
+      pattern: "*.gfa"
+  - gv:
+      type: file
+      description: Repeat graph
+      pattern: "*.gv"
+  - txt:
+      type: file
+      description: Extra contig information
+      pattern: "*.txt"
+  - log:
+      type: file
+      description: Flye log file
+      pattern: "*.log"
+  - json:
+      type: file
+      description: Flye parameters
+      pattern: "*.json"
+
+authors:
+  - "@asthara10"

tests/config/pytest_modules.yml

@@ -406,6 +406,10 @@ flash:
   - modules/flash/**
   - tests/modules/flash/**
 
+flye:
+  - modules/flye/**
+  - tests/modules/flye/**
+
 freebayes:
   - modules/freebayes/**
   - tests/modules/freebayes/**

tests/modules/flye/main.nf

@@ -0,0 +1,53 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { FLYE } from '../../../modules/flye/main.nf' addParams( options: [:] )
+
+workflow test_flye_pacbio_raw {
+    input = [ [ id:'test', mode:'pacbio-raw' ], // meta map
+              [ file(params.test_data['bacteroides_fragilis']['nanopore']['test_fastq_gz'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}
+
+workflow test_flye_pacbio_corr {
+    input = [ [ id:'test', mode:'pacbio-corr' ], // meta map
+              [ file(params.test_data['homo_sapiens']['pacbio']['hifi'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}
+
+workflow test_flye_pacbio_hifi {
+    input = [ [ id:'test', mode:'pacbio-hifi' ], // meta map
+              [ file(params.test_data['homo_sapiens']['pacbio']['hifi'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}
+
+workflow test_flye_nano_raw {
+    input = [ [ id:'test', mode:'nano-raw' ], // meta map
+              [ file(params.test_data['bacteroides_fragilis']['nanopore']['test_fastq_gz'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}
+
+workflow test_flye_nano_corr {
+    input = [ [ id:'test', mode:'nano-corr' ], // meta map
+              [ file(params.test_data['homo_sapiens']['pacbio']['hifi'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}
+
+workflow test_flye_nano_hq {
+    input = [ [ id:'test', mode:'nano-hq' ], // meta map
+              [ file(params.test_data['homo_sapiens']['pacbio']['hifi'], checkIfExists: true) ]
+            ]
+
+    FLYE ( input )
+}

tests/modules/flye/test.yml

@@ -0,0 +1,101 @@
+- name: flye test_flye_pacbio_raw
+  command: nextflow run tests/modules/flye -entry test_flye_pacbio_raw -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: 6f71c1083c4f992b5b4228036e2ff3c0
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: 4deb890fc3bf59a9dc3fdcd750dcbd5d
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: 0aa137b7bced1574f44196a297bbd1a1
+    - path: output/flye/test.assembly_info.txt
+      md5sum: 68eaa8e0cdf6c4c4ac855c510db8c71f
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9
+
+- name: flye test_flye_pacbio_corr
+  command: nextflow run tests/modules/flye -entry test_flye_pacbio_corr -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: 819d6fb836fa673ddf7a467e5a848b6f
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: ec582080d03db6c171a76db36eb26f10
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: 73399f4bce4c48ceda120a845dbb3c8c
+    - path: output/flye/test.assembly_info.txt
+      md5sum: a1695c9054cdff5f7a32f9a5294f2ffb
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9
+
+- name: flye test_flye_pacbio_hifi
+  command: nextflow run tests/modules/flye -entry test_flye_pacbio_hifi -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: a43006081eef11e5f8d0ccf23683364d
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: be04a4a6c55199cbd523b3acdd63ce50
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: 22e124521d4eb601ebe799735701f2d2
+    - path: output/flye/test.assembly_info.txt
+      md5sum: 506742757032551329626327b332cec8
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9
+
+- name: flye test_flye_nano_raw
+  command: nextflow run tests/modules/flye -entry test_flye_nano_raw -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: 423606b50dfdf41e59308ba795e7db71
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: 302ac6264e67925a22d3653d38ccb9eb
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: b18bd15f95607500d325ea23cf23ab50
+    - path: output/flye/test.assembly_info.txt
+      md5sum: 8123dbbfaeb840821302d620831ee489
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9
+
+- name: flye test_flye_nano_corr
+  command: nextflow run tests/modules/flye -entry test_flye_nano_corr -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: eaad569928775fee0025903670f2084d
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: ab7c0897ab4adcaef345c6d5bf560053
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: cbf5dfb40b25ba19faf64c315c23dbc1
+    - path: output/flye/test.assembly_info.txt
+      md5sum: 9791a3af96936b69910dc043dbb3ac27
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9
+
+- name: flye test_flye_nano_hq
+  command: nextflow run tests/modules/flye -entry test_flye_nano_hq -c tests/config/nextflow.config
+  tags:
+    - flye
+  files:
+    - path: output/flye/test.assembly.fasta
+      md5sum: 069936228fd00330b689c6638fc68fc2
+    - path: output/flye/test.assembly_graph.gfa
+      md5sum: 56e3a79f08303d2aea3533c0c11b8398
+    - path: output/flye/test.assembly_graph.gv
+      md5sum: 4dff1ebda5884cf369a6ae902029c52b
+    - path: output/flye/test.assembly_info.txt
+      md5sum: 26f699b25ab84ef7b1bf48c89e04e353
+    - path: output/flye/test.flye.log
+    - path: output/flye/test.params.json
+      md5sum: 54b576cb6d4d27656878a7fd3657bde9