Cooler cload (#634)

* rebuild cooler cload.

* update test file path of cload.

* add pytest for cload

* update to version.yml

* update the test data path

* Update tests/modules/cooler/cload/main.nf

Co-authored-by: Sébastien Guizard <sguizard@ed.ac.uk>

* Update modules/cooler/cload/main.nf

Co-authored-by: Sébastien Guizard <sguizard@ed.ac.uk>

* Update tests/config/test_data.config

Co-authored-by: Sébastien Guizard <sguizard@ed.ac.uk>

* Update main.nf

Remove a lonely curly bracket.

* Update test.yml

Updated with new workflows.

* update the test files

* merge the conflicts.

* update the test.yml

* update for change of cooler/dump

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

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+    // Do not publish versions.yml unless running from pytest workflow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

@@ -0,0 +1,47 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName; getProcessName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process COOLER_CLOAD {
+    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(pairs), path(index)
+    val cool_bin
+    path chromsizes
+
+    output:
+    tuple val(meta), val(cool_bin), path("*.cool"), emit: cool
+    path "versions.yml"                           , emit: versions
+
+    script:
+    def prefix = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def nproc  = options.args.contains('pairix') || options.args.contains('tabix')? "--nproc ${task.cpus}" : ''
+
+    """
+    cooler cload \\
+        $options.args \\
+        $nproc \\
+        ${chromsizes}:${cool_bin} \\
+        $pairs \\
+        ${prefix}.${cool_bin}.cool
+
+    cat <<-END_VERSIONS > versions.yml
+    ${getProcessName(task.process)}:
+        ${getSoftwareName(task.process)}: \$(cooler --version 2>&1 | sed 's/cooler, version //')
+    END_VERSIONS
+    """
+}

modules/cooler/cload/meta.yml

@@ -0,0 +1,52 @@
+name: cooler_cload
+description: Create a cooler from genomic pairs and bins
+keywords:
+  - cool
+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 ]
+  - pairs:
+      type: file
+      description: Path to contacts (i.e. read pairs) file.
+  - index:
+      type: file
+      description: Path to index file of the contacts.
+  - cool_bin:
+      type: value
+      description: Bins size in bp
+  - chromsizes:
+      type: file
+      description: Path to a chromsizes file.
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - version:
+      type: file
+      description: File containing software version
+      pattern: "versions.yml"
+  - cool:
+      type: file
+      description: Output COOL file path
+      pattern: "*.cool"
+  - cool_bin:
+      type: value
+      description: Bins size in bp
+
+authors:
+  - "@jianhong"

tests/config/pytest_modules.yml

@@ -298,6 +298,10 @@ cooler/digest:
   - modules/cooler/digest/**
   - tests/modules/cooler/digest/**
 
+cooler/cload:
+  - modules/cooler/cload/**
+  - tests/modules/cooler/cload/**
+  
 cooler/dump:
   - modules/cooler/dump/**
   - tests/modules/cooler/dump/**

tests/config/test_data.config

@@ -263,7 +263,13 @@ params {
             'txt' {
                 hello = "${test_data_dir}/generic/txt/hello.txt"
             }
-            'cooler' {
+            'cooler'{
+                test_pairix_pair_gz                           = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.blksrt.txt.gz"
+                test_pairix_pair_gz_px2                       = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.blksrt.txt.gz.px2"
+                test_pairs_pair                               = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.sample1.pairs"
+                test_tabix_pair_gz                            = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.sorted.possrt.txt.gz"
+                test_tabix_pair_gz_tbi                        = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.GM12878-MboI.pairs.subsample.sorted.possrt.txt.gz.tbi"
+                hg19_chrom_sizes                              = "${test_data_dir}/genomics/homo_sapiens/cooler/cload/hg19/hg19.chrom.sizes"
                 test_merge_cool                               = "${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"
 

tests/modules/cooler/cload/main.nf

@@ -0,0 +1,52 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { COOLER_CLOAD }                       from '../../../../modules/cooler/cload/main.nf' addParams( options: [args:'pairix'] )
+include { COOLER_CLOAD as COOLER_CLOAD_PAIRS } from '../../../../modules/cooler/cload/main.nf' addParams( options: [args:'pairs --chrom1 1 --pos1 2 --chrom2 4 --pos2 5 -N'] )
+include { COOLER_CLOAD as COOLER_CLOAD_TABIX } from '../../../../modules/cooler/cload/main.nf' addParams( options: [args:'tabix'] )
+include { COOLER_DUMP }                        from '../../../../modules/cooler/dump/main.nf' addParams( options: [:] )
+include { COOLER_DUMP as COOLER_DUMP_PAIRS}    from '../../../../modules/cooler/dump/main.nf' addParams( options: [:] )
+include { COOLER_DUMP as COOLER_DUMP_TABIX}    from '../../../../modules/cooler/dump/main.nf' addParams( options: [:] )
+
+workflow test_cooler_cload_pairix {
+
+    input = [ [ id:'test_pairix', single_end:false ], // meta map
+             file(params.test_data['generic']['cooler']['test_pairix_pair_gz'], checkIfExists: true),
+             file(params.test_data['generic']['cooler']['test_pairix_pair_gz_px2'], checkIfExists: true)]
+
+    sizes    = file(params.test_data['generic']['cooler']['hg19_chrom_sizes'], checkIfExists: true)
+    bin_size = 2000000
+
+    COOLER_CLOAD ( input, bin_size, sizes )
+    COOLER_DUMP(COOLER_CLOAD.out.cool.map{[it[0], it[2]]}, [])
+
+}
+
+workflow test_cooler_cload_pairs {
+
+    input = [ [ id:'test_pairs', single_end:false ], // meta map
+             file(params.test_data['generic']['cooler']['test_pairs_pair'], checkIfExists: true),
+             []]
+
+    sizes    = file(params.test_data['generic']['cooler']['hg19_chrom_sizes'], checkIfExists: true)
+    bin_size = 2000000
+
+    COOLER_CLOAD_PAIRS ( input, bin_size, sizes )
+    COOLER_DUMP_PAIRS(COOLER_CLOAD_PAIRS.out.cool.map{[it[0], it[2]]}, [])
+
+}
+
+workflow test_cooler_cload_tabix {
+
+    input = [ [ id:'test_tabix', single_end:false ], // meta map
+             file(params.test_data['generic']['cooler']['test_tabix_pair_gz'], checkIfExists: true),
+             file(params.test_data['generic']['cooler']['test_tabix_pair_gz_tbi'], checkIfExists: true)]
+
+    sizes    = file(params.test_data['generic']['cooler']['hg19_chrom_sizes'], checkIfExists: true)
+    bin_size = 2000000
+
+    COOLER_CLOAD_TABIX ( input, bin_size, sizes )
+    COOLER_DUMP_TABIX(COOLER_CLOAD_TABIX.out.cool.map{[it[0], it[2]]}, [])
+
+}

tests/modules/cooler/cload/test.yml

@@ -0,0 +1,29 @@
+- name: cooler cload test_cooler_cload_pairix
+  command: nextflow run tests/modules/cooler/cload -entry test_cooler_cload_pairix -c tests/config/nextflow.config
+  tags:
+    - cooler/cload
+    - cooler
+  files:
+    - path: output/cooler/test_pairix.2000000.cool
+    - path: output/cooler/test_pairix.bedpe
+      md5sum: 0cd85311089669688ec17468eae02111
+
+- name: cooler cload test_cooler_cload_pairs
+  command: nextflow run tests/modules/cooler/cload -entry test_cooler_cload_pairs -c tests/config/nextflow.config
+  tags:
+    - cooler/cload
+    - cooler
+  files:
+    - path: output/cooler/test_pairs.2000000.cool
+    - path: output/cooler/test_pairs.bedpe
+      md5sum: 7f832733fc7853ebb1937b33e4c1e0de
+
+- name: cooler cload test_cooler_cload_tabix
+  command: nextflow run tests/modules/cooler/cload -entry test_cooler_cload_tabix -c tests/config/nextflow.config
+  tags:
+    - cooler/cload
+    - cooler
+  files:
+    - path: output/cooler/test_tabix.2000000.cool
+    - path: output/cooler/test_tabix.bedpe
+      md5sum: 0cd85311089669688ec17468eae02111