add qcat module (from nanoseq modules) (#469)

* add qcat module

* remove md5sum(nom-reproducible)

software/qcat/functions.nf

@@ -0,0 +1,70 @@
+/*
+ * -----------------------------------------------------
+ *  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/qcat/main.nf

@@ -0,0 +1,44 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+def options    = initOptions(params.options)
+
+process QCAT {
+    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::qcat=1.1.0" : null)
+    container "quay.io/biocontainers/qcat:1.1.0--py_0"
+
+    input:
+    tuple val(meta), path(input_path)
+    val(barcode_kit)
+
+    output:
+    tuple val(meta), path("fastq/*.fastq.gz") , emit: fastq
+    path "*.version.txt"                      , emit: version
+
+    script:
+    """
+    ## Unzip fastq file
+    ## qcat doesnt support zipped files yet
+    FILE=$input_path
+    if [[ \$FILE == *.gz ]]
+    then
+    zcat $input_path > unzipped.fastq
+    FILE=unzipped.fastq
+    fi
+    qcat  \\
+    -f \$FILE \\
+    -b ./fastq \\
+    --kit $barcode_kit
+
+    ## Zip fastq files (cannot find pigz command)
+    gzip fastq/*
+    qcat --version &> qcat.version.txt
+    """
+}

software/qcat/meta.yml

@@ -0,0 +1,38 @@
+name: qcat
+description: Demultiplexer for Nanopore samples
+keywords:
+    - demultiplex
+
+tools:
+    - qcat:
+        description: |
+            A demultiplexer for Nanopore samples
+        homepage: https://github.com/nanoporetech/qcat
+        documentation: https://github.com/nanoporetech/qcat#qcat
+input:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - input_path:
+        type: file
+        description: |
+            Non-demultiplexed Nanopore sequencing sample.
+
+output:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - fastq:
+        type: file
+        description: Demultiplexed fastq samples
+        pattern: "*.fastq.gz"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@yuukiiwa"

tests/config/pytest_software.yml

@@ -439,6 +439,10 @@ pycoqc:
   - software/pycoqc/**
   - tests/software/pycoqc/**
 
+qcat:
+  - software/qcat/**
+  - tests/software/qcat/**
+
 qualimap/bamqc:
   - software/qualimap/bamqc/**
   - tests/software/qualimap/bamqc/**

tests/config/test_data.config

@@ -146,6 +146,9 @@ params {
                 test2_genome_vcf_gz_tbi                        = "${test_data_dir}/genomics/homo_sapiens/illumina/gvcf/test2.genome.vcf.gz.tbi"
                 test2_genome_vcf_idx                           = "${test_data_dir}/genomics/homo_sapiens/illumina/gvcf/test2.genome.vcf.idx"
             }
+            'nanopore' {
+                non_demultiplexed_fastq                        = "https://github.com/nf-core/test-datasets/raw/nanoseq/fastq/nondemultiplexed/sample_nobc_dx.fastq.gz"
+            }
         }
     }
 }

tests/software/qcat/main.nf

@@ -0,0 +1,14 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { QCAT  } from '../../../software/qcat/main.nf' addParams( options: [:] )
+
+workflow test_qcat {
+    def input = []
+    input = [ [ id:'test' ], // meta map
+            [ file(params.test_data['homo_sapiens']['nanopore']['non_demultiplexed_fastq'], checkIfExists: true) ]
+            ]
+    barcode_kit = 'NBD103/NBD104'
+    QCAT ( input, barcode_kit )
+}

tests/software/qcat/test.yml

@@ -0,0 +1,8 @@
+- name: qcat
+  command: nextflow run ./tests/software/qcat -entry test_qcat -c tests/config/nextflow.config
+  tags:
+    - qcat
+  files:
+    - path: ./output/qcat/fastq/barcode06.fastq.gz
+    - path: ./output/qcat/fastq/barcode12.fastq.gz
+    - path: ./output/qcat/fastq/none.fastq.gz