Cooler zoomify (#514)

* add software/cooler

* fix the wrong files uploaded.

* create a branch for cooler/zoomify

* Apply suggestions from code review

* update functions.nf to new version.

* update the test file to test-datasets.

* update the test method of zoomify

* update dump test file.

* update version.txt to version.yml

* Update modules/cooler/dump/main.nf

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

* fix the output bug of versions update to pytest_modules.yml

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

* Update modules/cooler/dump/main.nf

* indent

Co-authored-by: Harshil Patel <drpatelh@users.noreply.github.com>
Co-authored-by: Gregor Sturm <mail@gregor-sturm.de>
Co-authored-by: Sébastien Guizard <sguizard@ed.ac.uk>
Co-authored-by: FriederikeHanssen <Friederike.hanssen@qbic.uni-tuebingen.de>

modules/cooler/dump/main.nf

@@ -20,6 +20,7 @@ process COOLER_DUMP {
 
     input:
     tuple val(meta), path(cool)
+    val resolution
 
     output:
     tuple val(meta), path("*.bedpe"), emit: bedpe
@@ -27,11 +28,12 @@ process COOLER_DUMP {
 
     script:
     def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def suffix   = resolution     ? "::$resolution"               : ""
     """
     cooler dump \\
         $options.args \\
         -o ${prefix}.bedpe \\
-        $cool
+        $cool$suffix
 
     cat <<-END_VERSIONS > versions.yml
     ${getProcessName(task.process)}:

modules/cooler/zoomify/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/zoomify/main.nf

@@ -0,0 +1,42 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process COOLER_ZOOMIFY {
+    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("*.mcool"), emit: mcool
+    path "versions.yml"             , emit: versions
+
+    script:
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    cooler zoomify \\
+        $options.args \\
+        -n $task.cpus \\
+        -o ${prefix}.mcool \\
+        $cool
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$(cooler --version 2>&1 | sed 's/cooler, version //')
+    END_VERSIONS
+    """
+}

modules/cooler/zoomify/meta.yml

@@ -0,0 +1,41 @@
+name: cooler_zoomify
+description: Generate a multi-resolution cooler file by coarsening
+keywords:
+  - mcool
+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 versions
+      pattern: "versions.yml"
+  - mcool:
+      type: file
+      description: Output mcool file
+      pattern: "*.mcool"
+
+authors:
+  - "@jianhong"

tests/config/pytest_modules.yml

@@ -294,6 +294,10 @@ cooler/dump:
   - modules/cooler/dump/**
   - tests/modules/cooler/dump/**
 
+cooler/zoomify:
+  - modules/cooler/zoomify/**
+  - tests/software/cooler/zoomify/**
+
 cooler/merge:
   - modules/cooler/merge/**
   - tests/modules/cooler/merge/**

tests/config/test_data.config

@@ -260,6 +260,7 @@ params {
             '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'{

tests/modules/cooler/dump/main.nf

@@ -9,5 +9,5 @@ workflow test_cooler_dump {
     input = [ [ id:'test' ], // meta map
               file("https://raw.githubusercontent.com/open2c/cooler/master/tests/data/toy.asymm.16.cool", checkIfExists: true) ]
 
-    COOLER_DUMP ( input )
+    COOLER_DUMP ( input, [:] )
 }

tests/modules/cooler/zoomify/main.nf

@@ -0,0 +1,14 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { COOLER_ZOOMIFY } from '../../../../modules/cooler/zoomify/main.nf' addParams( options: ['args':'-r 2,4,8', publish_files:[:]] )
+include { COOLER_DUMP } from '../../../../modules/cooler/dump/main.nf' addParams( options: [:] )
+
+workflow test_cooler_zoomify {
+   input = [ [ id:'test' ], // meta map
+            file(params.test_data['generic']['cooler']['test_merge_cool'], checkIfExists: true)]
+
+    COOLER_ZOOMIFY ( input )
+    COOLER_DUMP(COOLER_ZOOMIFY.out.mcool, "/resolutions/2")
+}

tests/modules/cooler/zoomify/test.yml

@@ -0,0 +1,8 @@
+- name: cooler zoomify test_cooler_zoomify
+  command: nextflow run tests/modules/cooler/zoomify -entry test_cooler_zoomify -c tests/config/nextflow.config
+  tags:
+    - cooler
+    - cooler/zoomify
+  files:
+    - path: output/cooler/test.bedpe
+      md5sum: 8d792beb609fff62b536c326661f9507