Merge pull request #136 from nf-core/metaphlan3/mergemetaphlantables

Add MetaPhlAn3 per-database multi-sample taxon tables
2024-11-22 14:59:54 +00:00 · 2022-09-19 13:45:50 +02:00 · 2022-09-19 13:45:50 +02:00 · 187946ec80
commit 187946ec80
parent a2a9f458ec 016efcb807
8 changed files with 115 additions and 9 deletions
--- a/conf/modules.config
+++ b/conf/modules.config
@ -336,7 +336,7 @@ process {
        ]
    }

-    withName: METAPHLAN3 {
+    withName: METAPHLAN3_METAPHLAN3 {
        ext.args = { "${meta.db_params}" }
        ext.prefix = params.perform_runmerging ? { "${meta.id}-${meta.db_name}" } : { "${meta.id}-${meta.run_accession}-${meta.db_name}" }
        publishDir = [
@ -346,6 +346,15 @@ process {
        ]
    }

+    withName: METAPHLAN3_MERGEMETAPHLANTABLES {
+        ext.prefix = { "metaphlan3_${meta.id}_combined_reports" }
+        publishDir = [
+            path: { "${params.outdir}/metaphlan3/" },
+            mode: params.publish_dir_mode,
+            pattern: '*.{txt}'
+        ]
+    }
+
    withName: CENTRIFUGE_CENTRIFUGE {
        publishDir = [
            path: { "${params.outdir}/centrifuge/${meta.db_name}/" },
--- a/modules.json
+++ b/modules.json
@ -97,9 +97,13 @@
                    "branch": "master",
                    "git_sha": "2d38566eca4cc15142b2ffa7c11837569b39aece"
                },
+                "metaphlan3/mergemetaphlantables": {
+                    "branch": "master",
+                    "git_sha": "36bcd675ae76a379a38165898a203f4915823f4f"
+                },
                "metaphlan3/metaphlan3": {
                    "branch": "master",
-                    "git_sha": "940d7fe9d63962e0e2ba0987e2893fb0aff832e3"
+                    "git_sha": "978087354eb72ac1f6e18a3f790fad9bc4d05840"
                },
                "minimap2/align": {
                    "branch": "master",
--- a/modules/nf-core/modules/metaphlan3/mergemetaphlantables/main.nf
+++ b/modules/nf-core/modules/metaphlan3/mergemetaphlantables/main.nf
@ -0,0 +1,33 @@
+process METAPHLAN3_MERGEMETAPHLANTABLES {
+    label 'process_single'
+
+    conda (params.enable_conda ? 'bioconda::metaphlan=3.0.12' : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/metaphlan:3.0.12--pyhb7b1952_0' :
+        'quay.io/biocontainers/metaphlan:3.0.12--pyhb7b1952_0' }"
+
+    input:
+    tuple val(meta), path(profiles)
+
+    output:
+    tuple val(meta), path("${prefix}.txt") , emit: txt
+    path "versions.yml" , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    merge_metaphlan_tables.py \\
+        $args \\
+        -o ${prefix}.txt \\
+        ${profiles}
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        metaphlan3: \$(metaphlan --version 2>&1 | awk '{print \$3}')
+    END_VERSIONS
+    """
+}
--- a/modules/nf-core/modules/metaphlan3/mergemetaphlantables/meta.yml
+++ b/modules/nf-core/modules/metaphlan3/mergemetaphlantables/meta.yml
@ -0,0 +1,44 @@
+name: "metaphlan3_mergemetaphlantables"
+description: Merges output abundance tables from MetaPhlAn3
+keywords:
+  - metagenomics
+  - classification
+  - merge
+  - table
+  - profiles
+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 ]
+  - profiles:
+      type: file
+      description: List of per-sample MetaPhlAn3 taxonomic abundance tables
+      pattern: "*"
+
+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"
+  - txt:
+      type: txt
+      description: Combined MetaPhlAn3 table
+      pattern: "*.txt"
+
+authors:
+  - "@jfy133"
--- a/modules/nf-core/modules/metaphlan3/metaphlan3/main.nf
+++ b/modules/nf-core/modules/metaphlan3/metaphlan3/main.nf
@ -1,4 +1,4 @@
-process METAPHLAN3 {
+process METAPHLAN3_METAPHLAN3 {
    tag "$meta.id"
    label 'process_high'

--- a/modules/nf-core/modules/metaphlan3/metaphlan3/meta.yml
+++ b/modules/nf-core/modules/metaphlan3/metaphlan3/meta.yml
@ -1,4 +1,4 @@
-name: metaphlan3
+name: metaphlan3_metaphlan3
 description: MetaPhlAn is a tool for profiling the composition of microbial communities from metagenomic shotgun sequencing data.
 keywords:
  - metagenomics
--- a/subworkflows/local/profiling.nf
+++ b/subworkflows/local/profiling.nf
@ -7,7 +7,7 @@ include { MEGAN_RMA2INFO as MEGAN_RMA2INFO_TSV  } from '../../modules/nf-core/mo
 include { KRAKEN2_KRAKEN2                       } from '../../modules/nf-core/modules/kraken2/kraken2/main'
 include { CENTRIFUGE_CENTRIFUGE                 } from '../../modules/nf-core/modules/centrifuge/centrifuge/main'
 include { CENTRIFUGE_KREPORT                    } from '../../modules/nf-core/modules/centrifuge/kreport/main'
-include { METAPHLAN3                            } from '../../modules/nf-core/modules/metaphlan3/metaphlan3/main'
+include { METAPHLAN3_METAPHLAN3                 } from '../../modules/nf-core/modules/metaphlan3/metaphlan3/main'
 include { KAIJU_KAIJU                           } from '../../modules/nf-core/modules/kaiju/kaiju/main'
 include { DIAMOND_BLASTX                        } from '../../modules/nf-core/modules/diamond/blastx/main'
 include { MOTUS_PROFILE                         } from '../../modules/nf-core/modules/motus/profile/main'
@ -168,9 +168,9 @@ workflow PROFILING {
                                    db: it[3]
                            }

-        METAPHLAN3 ( ch_input_for_metaphlan3.reads, ch_input_for_metaphlan3.db )
-        ch_versions        = ch_versions.mix( METAPHLAN3.out.versions.first() )
-        ch_raw_profiles    = ch_raw_profiles.mix( METAPHLAN3.out.biom )
+        METAPHLAN3_METAPHLAN3 ( ch_input_for_metaphlan3.reads, ch_input_for_metaphlan3.db )
+        ch_versions        = ch_versions.mix( METAPHLAN3_METAPHLAN3.out.versions.first() )
+        ch_raw_profiles    = ch_raw_profiles.mix( METAPHLAN3_METAPHLAN3.out.profile )

    }

--- a/subworkflows/local/standardisation_profiles.nf
+++ b/subworkflows/local/standardisation_profiles.nf
@ -2,9 +2,10 @@
 // Standardise output files e.g. aggregation
 //

-include { KRAKENTOOLS_COMBINEKREPORTS as KRAKENTOOLS_COMBINEKREPORTS_CENTRIFUGE } from '../../modules/nf-core/modules/krakentools/combinekreports/main'
 include { KAIJU_KAIJU2TABLE                                                     } from '../../modules/nf-core/modules/kaiju/kaiju2table/main'
 include { KRAKENTOOLS_COMBINEKREPORTS                                           } from '../../modules/nf-core/modules/krakentools/combinekreports/main'
+include { KRAKENTOOLS_COMBINEKREPORTS as KRAKENTOOLS_COMBINEKREPORTS_CENTRIFUGE } from '../../modules/nf-core/modules/krakentools/combinekreports/main'
+include { METAPHLAN3_MERGEMETAPHLANTABLES                                       } from '../../modules/nf-core/modules/metaphlan3/mergemetaphlantables/main'
 include { MOTUS_MERGE                                                           } from '../../modules/nf-core/modules/motus/merge/main'

 workflow STANDARDISATION_PROFILES {
@ -27,6 +28,7 @@ workflow STANDARDISATION_PROFILES {
            motus: it[0]['tool'] == 'motus'
            kraken2: it[0]['tool'] == 'kraken2'
            centrifuge: it[0]['tool'] == 'centrifuge'
+            metaphlan3: it[0]['tool'] == 'metaphlan3'
            unknown: true
        }

@ -96,6 +98,20 @@ workflow STANDARDISATION_PROFILES {
    ch_multiqc_files = ch_multiqc_files.mix( KRAKENTOOLS_COMBINEKREPORTS.out.txt )
    ch_versions = ch_versions.mix( KRAKENTOOLS_COMBINEKREPORTS.out.versions )

+    // MetaPhlAn3
+    
+    ch_profiles_for_metaphlan3 = ch_input_profiles.metaphlan3
+                            .map { [it[0]['db_name'], it[1]] }
+                            .groupTuple()
+                            .map {
+                                [[id:it[0]], it[1]]
+                            }
+
+    METAPHLAN3_MERGEMETAPHLANTABLES ( ch_profiles_for_metaphlan3 )
+    ch_standardised_tables = ch_standardised_tables.mix( METAPHLAN3_MERGEMETAPHLANTABLES.out.txt )
+    ch_multiqc_files = ch_multiqc_files.mix( METAPHLAN3_MERGEMETAPHLANTABLES.out.txt )
+    ch_versions = ch_versions.mix( METAPHLAN3_MERGEMETAPHLANTABLES.out.versions )
+
    // mOTUs

    // mOTUs has a 'single' database, and cannot create custom ones.