Add metaphlan3 module (#543)

* Add metaphlan3 module * remove whitespace
2024-12-22 11:08:17 +00:00 · 2021-06-25 07:15:15 +01:00 · 2021-06-25 07:15:15 +01:00 · 4d711a1428
commit 4d711a1428
parent e0b7952b56
6 changed files with 356 additions and 0 deletions
--- a/software/metaphlan3/functions.nf
+++ b/software/metaphlan3/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)}/$args.filename"
+        }
+    }
+}
--- a/software/metaphlan3/main.nf
+++ b/software/metaphlan3/main.nf
@ -0,0 +1,50 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+options        = initOptions(params.options)
+
+process METAPHLAN3 {
+    tag "$meta.id"
+    label 'process_high'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['id']) }
+
+    conda (params.enable_conda ? "bioconda::metaphlan=3.0.10" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/metaphlan:3.0.10--pyhb7b1952_0"
+    } else {
+        container "quay.io/biocontainers/metaphlan:3.0.10--pyhb7b1952_0"
+    }
+
+    input:
+    tuple val(meta), path(input)
+    path metaphlan_db
+
+    output:
+    tuple val(meta), path("*_profile.txt")   ,                emit: profile
+    tuple val(meta), path("*.biom")          ,                emit: biom
+    tuple val(meta), path('*.bowtie2out.txt'), optional:true, emit: bt2out
+    path "*.version.txt"                     ,                emit: version
+
+    script:
+    def software    = getSoftwareName(task.process)
+    def prefix      = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def input_type  = ("$input".endsWith(".fastq.gz")) ? "--input_type fastq" :  ("$input".contains(".fasta")) ? "--input_type fasta" : ("$input".endsWith(".bowtie2out.txt")) ? "--input_type bowtie2out" : "--input_type sam"
+    def input_data  = ("$input_type".contains("fastq")) && !meta.single_end ? "${input[0]},${input[1]}" : "$input"
+    def bowtie2_out = "$input_type" == "--input_type bowtie2out" || "$input_type" == "--input_type sam" ? '' : "--bowtie2out ${prefix}.bowtie2out.txt"
+
+    """
+    metaphlan \\
+        --nproc $task.cpus \\
+        $input_type \\
+        $input_data \\
+        $options.args \\
+        $bowtie2_out \\
+        --bowtie2db ${metaphlan_db} \\
+        --biom ${prefix}.biom \\
+        --output_file ${prefix}_profile.txt
+    echo \$(metaphlan --version 2>&1) | awk '{print \$3}' > ${software}.version.txt
+    """
+}
--- a/software/metaphlan3/meta.yml
+++ b/software/metaphlan3/meta.yml
@ -0,0 +1,52 @@
+name: metaphlan3
+description: MetaPhlAn is a tool for profiling the composition of microbial communities from metagenomic shotgun sequencing data.
+keywords:
+  - metagenomics
+  - classification
+  - fastq
+  - bam
+  - fasta
+tools:
+  - metaphlan3:
+      description:  Identify clades (phyla to species) present in the metagenome obtained from a microbiome sample and their relative abundance
+      homepage: https://huttenhower.sph.harvard.edu/metaphlan/
+      documentation: https://github.com/biobakery/MetaPhlAn
+      doi: "10.7554/eLife.65088"
+      licence: ['MIT License']
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - input:
+      type: file
+      description: Metaphlan 3.0 can classify the metagenome from a variety of input data types, including FASTQ files (single-end and paired-end), FASTA, bowtie2-produced SAM files (produced from alignments to the MetaPHlAn marker database) and intermediate bowtie2 alignment files (bowtie2out)
+      pattern: "*.{fastq.gz, fasta, fasta.gz, sam, bowtie2out.txt}"
+
+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}"
+  - profile:
+      type: file
+      description: Tab-separated output file of the predicted taxon relative abundances
+      pattern: "*.{txt}"
+  - biom:
+      type: file
+      description: General-use format for representing biological sample by observation contingency tables
+      pattern: "*.{biom}"
+  - bowtie2out:
+      type: file
+      description: Intermediate Bowtie2 output produced from mapping the metagenome against the MetaPHlAn marker database ( not compatible with `bowtie2out` files generated with MetaPhlAn versions below 3 )
+      pattern: "*.{bowtie2out.txt}"
+
+authors:
+  - "@MGordon09"
--- a/tests/config/pytest_software.yml
+++ b/tests/config/pytest_software.yml
@ -422,6 +422,10 @@ mash/sketch:
  - software/mash/sketch/**
  - tests/software/mash/sketch/**

+metaphlan3:
+  - software/metaphlan3/**
+  - tests/software/metaphlan3/**
+
 methyldackel/extract:
  - software/methyldackel/extract/**
  - tests/software/methyldackel/extract/**
--- a/tests/software/metaphlan3/main.nf
+++ b/tests/software/metaphlan3/main.nf
@ -0,0 +1,59 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { UNTAR          } from '../../../software/untar/main.nf'          addParams( options: [:] )
+include { SAMTOOLS_VIEW  } from '../../../software/samtools/view/main.nf'  addParams( options: ['suffix': '.sam'] )
+include { METAPHLAN3     } from '../../../software/metaphlan3/main.nf'     addParams( options: [ 'args':'--index mpa_v30_CHOCOPhlAn_201901 --add_viruses --bt2_ps very-sensitive-local' ] )
+
+workflow test_metaphlan3_single_end {
+
+    input = [ [ id:'test', single_end:true ], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true) ]
+            ]
+
+    db    = channel.fromPath('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/metaphlan_database.tar.gz', type: 'dir', checkIfExists: true)
+
+    UNTAR ( db )
+    METAPHLAN3 ( input, UNTAR.out.untar )
+}
+
+workflow test_metaphlan3_paired_end {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true) ]
+            ]
+
+    db    = channel.fromPath('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/metaphlan_database.tar.gz', type: 'dir', checkIfExists: true)
+
+
+    UNTAR ( db )
+    METAPHLAN3 ( input, UNTAR.out.untar )
+}
+
+workflow test_metaphlan3_sam {
+
+    input = [ [ id:'test'], // meta map
+              [ file(params.test_data['sarscov2']['illumina']['test_single_end_bam'], checkIfExists: true) ]
+            ]
+
+    db    = channel.fromPath('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/metaphlan_database.tar.gz', type: 'dir', checkIfExists: true)
+
+
+    UNTAR ( db )
+    SAMTOOLS_VIEW ( input )
+    METAPHLAN3 ( SAMTOOLS_VIEW.out.bam, UNTAR.out.untar )
+}
+
+workflow test_metaphlan3_fasta {
+
+    input = [ [ id:'test', single_end:true], // meta map
+              [ file(params.test_data['sarscov2']['genome']['transcriptome_fasta'], checkIfExists: true) ]
+            ]
+
+    db    = channel.fromPath('https://raw.githubusercontent.com/nf-core/test-datasets/modules/data/delete_me/metaphlan_database.tar.gz', type: 'dir', checkIfExists: true)
+
+    UNTAR ( db )
+    METAPHLAN3 ( input, UNTAR.out.untar )
+}
--- a/tests/software/metaphlan3/test.yml
+++ b/tests/software/metaphlan3/test.yml
@ -0,0 +1,123 @@
+- name: metaphlan3 test_metaphlan3_single_end
+  command: nextflow run tests/software/metaphlan3 -entry test_metaphlan3_single_end -c tests/config/nextflow.config
+  tags:
+    - metaphlan3
+  files:
+    - path: output/metaphlan3/test.biom
+      contains:
+      - '"format": "Biological Observation Matrix 1.0.0","format_url": "http://biom-format.org","generated_by"'
+    - path: output/metaphlan3/test.bowtie2out.txt
+      md5sum: ef46a9c6a8ce9cae26fbfd5527116fd5
+    - path: output/metaphlan3/test_profile.txt
+      md5sum: 72d40ee2304c162f3c165e1e578ed152
+    - path: output/untar/metaphlan_database/mpa_latest
+      md5sum: b1337362f607000384563a56a6ff4790
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.1.bt2
+      md5sum: d52a98fe273742ade7c744b819a7c5c1
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.2.bt2
+      md5sum: b14cc7faf3a4fae792160c917aebfe03
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.3.bt2
+      md5sum: 1bca5df879f2c6fad0c54984d0651bfb
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.4.bt2
+      md5sum: 249ee1c672d52d50cee41cb94b6adc42
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.fasta
+      md5sum: d3efe201c9eb449e877ead36656abf5f
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.pkl
+      md5sum: b208bb15eaef50d91cc7d5e35a1518ee
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.1.bt2
+      md5sum: ed05be063280e8629193e52903b07591
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.2.bt2
+      md5sum: 1ca16b905abf657b88ca2bc12e7ad404
+
+- name: metaphlan3 test_metaphlan3_paired_end
+  command: nextflow run tests/software/metaphlan3 -entry test_metaphlan3_paired_end -c tests/config/nextflow.config
+  tags:
+    - metaphlan3
+  files:
+    - path: output/metaphlan3/test.biom
+      contains:
+      - '"format": "Biological Observation Matrix 1.0.0","format_url": "http://biom-format.org","generated_by"'
+    - path: output/metaphlan3/test.bowtie2out.txt
+      md5sum: ce11486fcc0e68fe7152867a3634e09a
+    - path: output/metaphlan3/test_profile.txt
+      md5sum: fcf99fec08ee00db6ef2c12fb93bc14b
+    - path: output/untar/metaphlan_database/mpa_latest
+      md5sum: b1337362f607000384563a56a6ff4790
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.1.bt2
+      md5sum: d52a98fe273742ade7c744b819a7c5c1
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.2.bt2
+      md5sum: b14cc7faf3a4fae792160c917aebfe03
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.3.bt2
+      md5sum: 1bca5df879f2c6fad0c54984d0651bfb
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.4.bt2
+      md5sum: 249ee1c672d52d50cee41cb94b6adc42
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.fasta
+      md5sum: d3efe201c9eb449e877ead36656abf5f
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.pkl
+      md5sum: b208bb15eaef50d91cc7d5e35a1518ee
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.1.bt2
+      md5sum: ed05be063280e8629193e52903b07591
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.2.bt2
+      md5sum: 1ca16b905abf657b88ca2bc12e7ad404
+
+- name: metaphlan3 test_metaphlan3_sam
+  command: nextflow run tests/software/metaphlan3 -entry test_metaphlan3_sam -c tests/config/nextflow.config
+  tags:
+    - metaphlan3
+  files:
+    - path: output/metaphlan3/test.biom
+      contains:
+      - '"format": "Biological Observation Matrix 1.0.0","format_url": "http://biom-format.org","generated_by"'
+    - path: output/metaphlan3/test_profile.txt
+      md5sum: e050d49f7df8a23617880ef9ed7745a0
+    - path: output/samtools/test.sam.bam
+      md5sum: 1d5be3c91979ead358e3980e3e7c9acf
+    - path: output/untar/metaphlan_database/mpa_latest
+      md5sum: b1337362f607000384563a56a6ff4790
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.1.bt2
+      md5sum: d52a98fe273742ade7c744b819a7c5c1
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.2.bt2
+      md5sum: b14cc7faf3a4fae792160c917aebfe03
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.3.bt2
+      md5sum: 1bca5df879f2c6fad0c54984d0651bfb
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.4.bt2
+      md5sum: 249ee1c672d52d50cee41cb94b6adc42
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.fasta
+      md5sum: d3efe201c9eb449e877ead36656abf5f
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.pkl
+      md5sum: b208bb15eaef50d91cc7d5e35a1518ee
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.1.bt2
+      md5sum: ed05be063280e8629193e52903b07591
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.2.bt2
+      md5sum: 1ca16b905abf657b88ca2bc12e7ad404
+
+- name: metaphlan3 test_metaphlan3_fasta
+  command: nextflow run tests/software/metaphlan3 -entry test_metaphlan3_fasta -c tests/config/nextflow.config
+  tags:
+    - metaphlan3
+  files:
+    - path: output/metaphlan3/test.biom
+      contains:
+      - '"format": "Biological Observation Matrix 1.0.0","format_url": "http://biom-format.org","generated_by"'
+    - path: output/metaphlan3/test.bowtie2out.txt
+      md5sum: fece494a410b8328608a11de10af6396
+    - path: output/metaphlan3/test_profile.txt
+      md5sum: d148c4203e5a9d59f5eea57207e40769
+    - path: output/untar/metaphlan_database/mpa_latest
+      md5sum: b1337362f607000384563a56a6ff4790
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.1.bt2
+      md5sum: d52a98fe273742ade7c744b819a7c5c1
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.2.bt2
+      md5sum: b14cc7faf3a4fae792160c917aebfe03
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.3.bt2
+      md5sum: 1bca5df879f2c6fad0c54984d0651bfb
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.4.bt2
+      md5sum: 249ee1c672d52d50cee41cb94b6adc42
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.fasta
+      md5sum: d3efe201c9eb449e877ead36656abf5f
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.pkl
+      md5sum: b208bb15eaef50d91cc7d5e35a1518ee
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.1.bt2
+      md5sum: ed05be063280e8629193e52903b07591
+    - path: output/untar/metaphlan_database/mpa_v30_CHOCOPhlAn_201901.rev.2.bt2
+      md5sum: 1ca16b905abf657b88ca2bc12e7ad404