Cooler merge (#515)

* add software/cooler

* fix the wrong files uploaded.

* create a branch for cooler/merge

* remove the bin_size from metadata.

* update the test_data to test-datasets

* update pytest_modules.yml

* update the test file from single input file to two input file.
update the output file from hdf5 to bedpe.

* update the version.txt to version.yml and functions.nf

* change version.yml to versions

* update the test file path and fix the output versions.

* Update meta.yml

Correct "version" to "versions"

* Update main.nf

Fix typo

* Update main.nf

Remove some spaces

Co-authored-by: Gregor Sturm <mail@gregor-sturm.de>
Co-authored-by: Sébastien Guizard <sguizard@ed.ac.uk>

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+    // Do not publish versions.yml unless running from pytest workflow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

@@ -0,0 +1,41 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process COOLER_MERGE {
+    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 ? "bioconda::cooler=0.8.11" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/cooler:0.8.11--pyh3252c3a_0"
+    } else {
+        container "quay.io/biocontainers/cooler:0.8.11--pyh3252c3a_0"
+    }
+
+    input:
+    tuple val(meta), path(cool)
+
+    output:
+    tuple val(meta), path("*.cool"), emit: cool
+    path "versions.yml"            , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    cooler merge \\
+        $options.args \\
+        ${prefix}.cool \\
+        ${cool}
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$(cooler --version 2>&1 | sed 's/cooler, version //')
+    END_VERSIONS
+    """
+}

modules/cooler/merge/meta.yml

@@ -0,0 +1,41 @@
+name: cooler_merge
+description: Merge multiple coolers with identical axes
+keywords:
+  - merge
+tools:
+  - cooler:
+      description: Sparse binary format for genomic interaction matrices
+      homepage: https://cooler.readthedocs.io/en/latest/index.html
+      documentation: https://cooler.readthedocs.io/en/latest/index.html
+      tool_dev_url: https://github.com/open2c/cooler
+      doi: "10.1093/bioinformatics/btz540"
+      licence: ['BSD-3-clause']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - cool:
+      type: file
+      description: Path to COOL file
+      pattern: "*.{cool,mcool}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - versions:
+      type: file
+      description: File containing software version
+      pattern: "versions.yml"
+  - cool:
+      type: file
+      description: Path to COOL file
+      pattern: "*.cool"
+
+authors:
+  - "@jianhong"

tests/config/pytest_modules.yml

@@ -294,6 +294,10 @@ cooler/dump:
   - modules/cooler/dump/**
   - tests/modules/cooler/dump/**
 
+cooler/merge:
+  - modules/cooler/merge/**
+  - tests/modules/cooler/merge/**
+
 csvtk/concat:
   - modules/csvtk/concat/**
   - tests/modules/csvtk/concat/**

tests/config/test_data.config

@@ -255,6 +255,10 @@ params {
             'txt' {
                 hello = "${test_data_dir}/generic/txt/hello.txt"
             }
+            'cooler' {
+                test_merge_cool                               = "${test_data_dir}/genomics/homo_sapiens/cooler/merge/toy/toy.symm.upper.2.cool"
+                test_merge_cool_cp2                           = "${test_data_dir}/genomics/homo_sapiens/cooler/merge/toy/toy.symm.upper.2.cp2.cool"
+            }
         }
         'bacteroides_fragilis'{
             'genome' {

tests/modules/cooler/merge/main.nf

@@ -0,0 +1,16 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { COOLER_MERGE } from '../../../../modules/cooler/merge/main.nf' addParams( options: [publish_files:[:]] )
+include { COOLER_DUMP } from '../../../../modules/cooler/dump/main.nf' addParams( options: [:] )
+
+workflow test_cooler_merge {
+
+    input = [ [ id:'test' ], // meta map
+              [ file(params.test_data['generic']['cooler']['test_merge_cool'], checkIfExists: true),
+                file(params.test_data['generic']['cooler']['test_merge_cool_cp2'], checkIfExists: true)] 
+            ]
+
+    COOLER_MERGE ( input ).cool | COOLER_DUMP
+}

tests/modules/cooler/merge/test.yml

@@ -0,0 +1,8 @@
+- name: cooler merge test_cooler_merge
+  command: nextflow run tests/modules/cooler/merge -entry test_cooler_merge -c tests/config/nextflow.config
+  tags:
+    - cooler/merge
+    - cooler
+  files:
+    - path: output/cooler/test.bedpe
+      md5sum: 0ce5e715bfc4674cdda02f2d7e7e3170