Pairtools dedup (#519)

* add software/pairtools

* create a branch for pairtools/dedup

* remove dedup from the filename.

software/pairtools/dedup/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"
+        }
+    }
+}

software/pairtools/dedup/main.nf

@@ -0,0 +1,41 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process PAIRTOOLS_DEDUP {
+    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::pairtools=0.3.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/pairtools:0.3.0--py37hb9c2fc3_5"
+    } else {
+        container "quay.io/biocontainers/pairtools:0.3.0--py37hb9c2fc3_5"
+    }
+
+    input:
+    tuple val(meta), path(input)
+
+    output:
+    tuple val(meta), path("*.pairs.gz")  , emit: pairs
+    tuple val(meta), path("*.pairs.stat"), emit: stat
+    path "*.version.txt"                 , emit: version
+
+    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    pairtools dedup \\
+        $options.args \\
+        -o ${prefix}.pairs.gz \\
+        --output-stats ${prefix}.pairs.stat \\
+        $input
+
+    echo \$(pairtools --version 2>&1) | sed 's/pairtools.*version //' > ${software}.version.txt
+    """
+}

software/pairtools/dedup/meta.yml

@@ -0,0 +1,44 @@
+name: pairtools_dedup
+description: Find and remove PCR/optical duplicates
+keywords:
+  - dedup
+tools:
+  - pairtools:
+      description: CLI tools to process mapped Hi-C data
+      homepage: http://pairtools.readthedocs.io/
+      documentation: http://pairtools.readthedocs.io/
+      tool_dev_url: https://github.com/mirnylab/pairtools
+      doi: ""
+      licence: ['MIT']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input:
+      type: file
+      description: pair 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: "*.{version.txt}"
+  - pairs:
+      type: file
+      description: Duplicates removed pairs
+      pattern: "*.{pairs.gz}"
+  - stat:
+      type: file
+      description: stats of the pairs
+      pattern: "*.{pairs.stat}"
+
+authors:
+  - "@jianhong"

tests/config/pytest_software.yml

@@ -455,9 +455,9 @@ optitype:
   - software/optitype/**
   - tests/software/optitype/**
 
-pairix:
+pairtools/dedup:
-  - software/pairix/**
+  - software/pairtools/dedup/**
-  - tests/software/pairix/**
+  - tests/software/pairtools/dedup/**
 
 pangolin:
   - software/pangolin/**

tests/software/pairtools/dedup/main.nf

@@ -0,0 +1,13 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { PAIRTOOLS_DEDUP } from '../../../../software/pairtools/dedup/main.nf' addParams( options: ['suffix':'.dedup'] )
+
+workflow test_pairtools_dedup {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file("https://raw.githubusercontent.com/open2c/pairtools/master/tests/data/mock.4dedup.pairsam", checkIfExists: true) ]
+
+    PAIRTOOLS_DEDUP ( input )
+}

tests/software/pairtools/dedup/test.yml

@@ -0,0 +1,10 @@
+- name: pairtools dedup test_pairtools_dedup
+  command: nextflow run tests/software/pairtools/dedup -entry test_pairtools_dedup -c tests/config/nextflow.config
+  tags:
+    - pairtools/dedup
+    - pairtools
+  files:
+    - path: output/pairtools/test.dedup.pairs.gz
+      md5sum: 785ffc9495731fd3d0f6f679a0a0e988
+    - path: output/pairtools/test.dedup.pairs.stat
+      md5sum: a554a846567cbcda66f070c76a9173bc