New module: galah (#3666)

* Add galah * NO DOI * fix versions * fix dashing version in conda * fix inputs/outputs * fix dashing version * Update model inputs, add tests * Update error message. * Update main.nf * Update test.yml * Update main.nf * Update modules/nf-core/galah/main.nf Co-authored-by: Adam Talbot <[email protected]> * Update main.nf * Update modules/nf-core/galah/main.nf Co-authored-by: Adam Talbot <[email protected]> * Update main.nf * Update test * Update main.nf * Update main.nf * fix tests --------- Co-authored-by: Adam Talbot <[email protected]>
nf-core · Aug 7, 2023 · 4375e5a · 4375e5a
1 parent b58a51f
commit 4375e5a
Show file tree

Hide file tree

Showing 6 changed files with 218 additions and 0 deletions.
diff --git a/modules/nf-core/galah/main.nf b/modules/nf-core/galah/main.nf
@@ -0,0 +1,59 @@
+process GALAH {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda "bioconda::galah=0.3.1"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/galah%3A0.3.1--h031d066_3':
+        'biocontainers/galah:0.3.1--h031d066_3' }"
+
+    input:
+    tuple val(meta), path(bins), path(qc_table), val(qc_format)
+
+    output:
+    tuple val(meta), path("*.tsv")                    , emit: tsv
+    tuple val(meta), path("${prefix}-dereplicated/*") , emit: dereplicated_bins
+    path "versions.yml"                               , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def qc_args = (qc_format == "checkm") ? "--checkm-tab-table ${qc_table}" : "--genome-info ${qc_table}"
+    def qc_input = qc_table ? qc_args : ""
+    def valid_qc_format = qc_format in ["checkm", "genome_info"]
+    if( qc_table && !valid_qc_format ) {
+        error "Invalid qc_format supplied! qc_format should be either 'checkm' or 'genome_info'."
+    }
+    """
+    mkdir ${prefix}-dereplicated
+
+    galah cluster \\
+        --threads ${task.cpus} \\
+        --genome-fasta-files ${bins} \\
+        ${qc_input} \\
+        --output-cluster-definition ${prefix}-dereplicated_bins.tsv \\
+        --output-representative-fasta-directory ${prefix}-dereplicated
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        galah: \$(galah --version | sed 's/galah //')
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    mkdir ${prefix}-dereplicated/
+    touch ${prefix}-dereplicated/test.fa
+    touch ${prefix}-dereplicated_bins.tsv
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        galah: \$(galah --version | sed 's/galah //')
+    END_VERSIONS
+    """
+}
diff --git a/modules/nf-core/galah/meta.yml b/modules/nf-core/galah/meta.yml
@@ -0,0 +1,63 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/modules/yaml-schema.json
+name: "galah"
+description: Cluster genome FASTA files by average nucleotide identity
+keywords:
+  - genomics
+  - cluster
+  - genome
+  - metagenomics
+tools:
+  - "galah":
+      description: "Galah aims to be a more scalable metagenome assembled genome (MAG) dereplication method."
+      homepage: "https://github.com/wwood/galah"
+      documentation: "https://github.com/wwood/galah"
+      tool_dev_url: "https://github.com/wwood/galah"
+      doi: "10.1111/NODOI"
+      licence: "['GPL v3']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. `[ id:'test', single_end:false ]`
+  - bins:
+      type: file
+      description: A list of fasta-formatted genomes for dereplication
+      pattern: "*.{fa,fna,fa.gz, etc}"
+  - qc_table:
+      type: file
+      description: |
+        (optional) Either a (CheckM)[https://nf-co.re/modules/checkm_lineagewf] summary TSV containing
+        information on the completeness and contamination of the input genomes (13 columns),
+        or a 3-column csv with the header `genome,completeness,contamination`.
+        In both cases the first column should contain the names of the input genome files,
+        minus the last file extension
+        (i.e. if the genome is gzipped, the genome name should retain the .fasta extension).
+      pattern: "*.{csv,tsv}"
+  - qc_format:
+      type: string
+      description: Defines the type if input table in `qc_table`, if specified.
+      pattern: "checkm|genome_info"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. `[ id:'test', single_end:false ]`
+  - tsv:
+      type: file
+      description: TSV file in the format `representative_genome` \t `member_genome`
+      pattern: "*.tsv"
+  - dereplicated_bins:
+      type: file
+      description: The representative genomes following dereplication by galah.
+      pattern: "*"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+authors:
+  - "@prototaxites"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
@@ -1247,6 +1247,10 @@ freyja/variants:
   - modules/nf-core/freyja/variants/**
   - tests/modules/nf-core/freyja/variants/**
 
+galah:
+  - modules/nf-core/galah/**
+  - tests/modules/nf-core/galah/**
+
 gamma/gamma:
   - modules/nf-core/gamma/gamma/**
   - tests/modules/nf-core/gamma/gamma/**

diff --git a/tests/modules/nf-core/galah/main.nf b/tests/modules/nf-core/galah/main.nf
@@ -0,0 +1,49 @@
+
+include { GALAH                       } from '../../../../modules/nf-core/galah/main.nf'
+include { BIOAWK as BIOAWK_CHECKM     } from '../../../../modules/nf-core/bioawk/main.nf'
+include { BIOAWK as BIOAWK_GENOMEINFO } from '../../../../modules/nf-core/bioawk/main.nf'
+include { GUNZIP                      } from '../../../../modules/nf-core/gunzip/main.nf'
+
+workflow test_galah {
+
+    input = [
+        [ id:'test' ], // meta map
+        [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+        file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)],
+        [],
+        []
+    ]
+
+    GALAH ( input )
+
+}
+
+workflow test_galah_genomeinfo {
+
+    genomeinfo = Channel.fromPath("https://raw.githubusercontent.com/nf-core/test-datasets/magmap/testdata/checkm.lineage_wf.qa_2.tsv", checkIfExists: true)
+        .map { file ->
+            [ [id: "genomeinfo"], file ]
+    }
+
+    BIOAWK_GENOMEINFO(genomeinfo)
+
+    GUNZIP(BIOAWK_GENOMEINFO.out.output)
+
+    ch_genomeinfo = GUNZIP.out.gunzip
+        .map { meta, tsv -> [tsv] }
+
+    input = Channel.of(
+        [
+            [ id:'test' ], // meta map
+            [file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCA_002688505.1_ASM268850v1_genomic.fna.gz", checkIfExists: true),
+            file("https://github.com/nf-core/test-datasets/raw/magmap/testdata/GCF_004296495.1_ASM429649v1_genomic.fna.gz", checkIfExists: true)]
+        ]
+        )
+    .combine(ch_genomeinfo)
+    .map {meta, bins, qc ->
+            [ meta, bins, qc, "genome_info" ]
+    }
+
+    GALAH ( input )
+
+}
diff --git a/tests/modules/nf-core/galah/nextflow.config b/tests/modules/nf-core/galah/nextflow.config
@@ -0,0 +1,13 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    // write a horrid awk to munge the checkm_qa tsv to a 3 column csv
+    // as files are gzipped, genome has to include extension "fna" as galah expects the genome name to be
+    // the file name minus the last extension
+    withName: BIOAWK_GENOMEINFO {
+        ext.args = '\'BEGIN {{FS="\t"; OFS=","}} NR==1 {print "genome","completeness","contamination"} NR>1 {print $1".fna",$6, $7}\''
+        ext.prefix = "genome_info.tsv"
+    }
+
+}
diff --git a/tests/modules/nf-core/galah/test.yml b/tests/modules/nf-core/galah/test.yml
@@ -0,0 +1,30 @@
+- name: galah test_galah
+  command: nextflow run ./tests/modules/nf-core/galah -entry test_galah -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/galah/nextflow.config
+  tags:
+    - galah
+  files:
+    - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
+      md5sum: 0747c48f6693a4fb03c7164c2f472326
+    - path: output/galah/test-dereplicated/GCF_004296495.1_ASM429649v1_genomic.fna.gz
+      md5sum: a8e9bac598df938f25e09418ff7214dd
+    - path: output/galah/test-dereplicated_bins.tsv
+      md5sum: d2f8a621bfa5794467f4fdd759e2bce7
+    - path: output/galah/versions.yml
+
+- name: galah test_galah_genomeinfo
+  command: nextflow run ./tests/modules/nf-core/galah -entry test_galah_genomeinfo -c ./tests/config/nextflow.config -c ./tests/modules/nf-core/galah/nextflow.config
+  tags:
+    - galah
+  files:
+    - path: output/bioawk/genome_info.tsv.gz
+    - path: output/bioawk/versions.yml
+    - path: output/galah/test-dereplicated/GCA_002688505.1_ASM268850v1_genomic.fna.gz
+      md5sum: 0747c48f6693a4fb03c7164c2f472326
+    - path: output/galah/test-dereplicated/GCF_004296495.1_ASM429649v1_genomic.fna.gz
+      md5sum: a8e9bac598df938f25e09418ff7214dd
+    - path: output/galah/test-dereplicated_bins.tsv
+      md5sum: d2f8a621bfa5794467f4fdd759e2bce7
+    - path: output/galah/versions.yml
+    - path: output/gunzip/genome_info.tsv
+      md5sum: f73b9131ab91ddb754725b94e5085955
+    - path: output/gunzip/versions.yml