CAT CAT (#722)

* bbmap/align done

* Tests for single end and prebuilt index

* Write bam file directly

* Forgot to use all cpus for bbmap

* Test md5sums

* Added pigz support

* Update modules/bbmap/align/meta.yml

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* process_medium and fastq

* cat/cat module

* Remove filter from CAT_CAT

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>
Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>

modules/cat/cat/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"
+        }
+    }
+}

modules/cat/cat/main.nf

@@ -0,0 +1,39 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process CAT_CAT {
+    label 'process_low'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:[:], publish_by_meta:[]) }
+
+    conda (params.enable_conda ? "conda-forge::pigz=2.3.4" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/pigz:2.3.4"
+    } else {
+        container "quay.io/biocontainers/pigz:2.3.4"
+    }
+
+    input:
+    path files
+
+    output:
+    path "file*"        , emit: file
+    path "*.version.txt", emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    cpus = Math.floor(task.cpus/2).toInteger()
+
+    // Use options.suffix if specified, otherwise .out; add .gz if first input file has it
+    suffix  = options.suffix ? "${options.suffix}" : ".out"
+    suffix += files[0].name =~ /\.gz/ ? '.gz' : ''
+
+    """
+    cat ${options.args} $files ${options.args2} > file${suffix}
+    cat --version | grep 'GNU coreutils' | sed 's/cat (GNU coreutils) //' > ${software}.version.txt
+    """
+}

modules/cat/cat/meta.yml

@@ -0,0 +1,33 @@
+name: cat_cat
+description: A module for concatenation of gzipped or uncompressed files, optionally filtering the output
+keywords:
+  - concatenate
+  - gzip
+  - filter
+tools:
+  - cat:
+      description: Just concatenation
+      homepage: None
+      documentation: None
+      tool_dev_url: None
+      doi: ""
+      licence: ""
+
+input:
+  - files:
+      type: file
+      description: Gzipped or not files
+      pattern: "*"
+
+output:
+  - version:
+      type: file
+      description: File containing version of the pigz software
+      pattern: "*.{version.txt}"
+  - bam:
+      type: file
+      description: Concatenated, optionally filtered, file, gzipped if input was, otherwise not
+      pattern: "file*"
+
+authors:
+  - "@erikrikarddaniel"

tests/config/pytest_modules.yml

@@ -230,6 +230,10 @@ bwameth/index:
   - modules/bwameth/index/**
   - tests/modules/bwameth/index/**
 
+cat/cat:
+  - modules/cat/cat/**
+  - tests/modules/cat/cat/**
+
 cat/fastq:
   - modules/cat/fastq/**
   - tests/modules/cat/fastq/**

tests/modules/cat/cat/main.nf

@@ -0,0 +1,46 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { CAT_CAT } from '../../../../modules/cat/cat/main.nf' addParams( options: [:] )
+include { CAT_CAT as CAT_CAT_SUFFIX } from '../../../../modules/cat/cat/main.nf' addParams( options: [suffix: ".fna"] )
+
+workflow test_cat_ungzipped {
+    
+    input = [
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true),
+        file(params.test_data['sarscov2']['genome']['genome_sizes'], checkIfExists: true)
+    ]
+
+    CAT_CAT ( input )
+}
+
+workflow test_cat_gzipped {
+    
+    input = [
+        file(params.test_data['sarscov2']['genome']['genome_gff3_gz'], checkIfExists: true),
+        file(params.test_data['sarscov2']['genome']['contigs_genome_maf_gz'], checkIfExists: true)
+    ]
+
+    CAT_CAT ( input )
+}
+
+workflow test_cat_ungzipped_fna {
+    
+    input = [
+        file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true),
+        file(params.test_data['sarscov2']['genome']['genome_sizes'], checkIfExists: true)
+    ]
+
+    CAT_CAT_SUFFIX ( input )
+}
+
+workflow test_cat_gzipped_fna {
+    
+    input = [
+        file(params.test_data['sarscov2']['genome']['genome_gff3_gz'], checkIfExists: true),
+        file(params.test_data['sarscov2']['genome']['contigs_genome_maf_gz'], checkIfExists: true)
+    ]
+
+    CAT_CAT_SUFFIX ( input )
+}

tests/modules/cat/cat/test.yml

@@ -0,0 +1,33 @@
+- name: cat ungzipped
+  command: nextflow run ./tests/modules/cat/cat -entry test_cat_ungzipped -c tests/config/nextflow.config
+  tags:
+    - cat
+    - cat/cat
+  files:
+    - path: output/cat/file.out
+      md5sum: f44b33a0e441ad58b2d3700270e2dbe2
+
+- name: cat gzipped
+  command: nextflow run ./tests/modules/cat/cat -entry test_cat_gzipped -c tests/config/nextflow.config
+  tags:
+    - cat
+    - cat/cat
+  files:
+    - path: output/cat/file.out.gz
+
+- name: cat ungzipped suffix
+  command: nextflow run ./tests/modules/cat/cat -entry test_cat_ungzipped_fna -c tests/config/nextflow.config
+  tags:
+    - cat
+    - cat/cat
+  files:
+    - path: output/cat/file.fna
+      md5sum: f44b33a0e441ad58b2d3700270e2dbe2
+
+- name: cat gzipped suffix
+  command: nextflow run ./tests/modules/cat/cat -entry test_cat_gzipped_fna -c tests/config/nextflow.config
+  tags:
+    - cat
+    - cat/cat
+  files:
+    - path: output/cat/file.fna.gz