nf-core · andersgs · Jun 15, 2022 · Jun 16, 2022 · Jun 16, 2022 · Jun 15, 2022
diff --git a/modules/goalign/cleanseqs/main.nf b/modules/goalign/cleanseqs/main.nf
@@ -0,0 +1,37 @@
+process GOALIGN_CLEANSEQS {
+    tag "goalign-cleanseqs:${aln}";
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::goalign=0.3.5" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/goalign:0.3.5--h65a6115_0':
+        'quay.io/biocontainers/goalign:0.3.5--h65a6115_0' }"
+
+    input:
+    path(aln)
+
+    output:
+    path("*.{fasta,fas,fa,phy}"), emit: aln
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = "${aln.getName()}_clean"
+    """
+    goalign \\
+        clean \\
+        seqs \\
+        -t ${task.cpus} \\
+        $args \\
+        -o ${prefix}.${aln.getExtension()} \\
+        -i $aln
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        goalign: \$(echo \$(goalign version))
+    END_VERSIONS
+    """
+}
diff --git a/modules/goalign/cleanseqs/meta.yml b/modules/goalign/cleanseqs/meta.yml
@@ -0,0 +1,32 @@
+name: "goalign_cleanseqs"
+description: Filter out poor sequences from a multisequence alignment
+keywords:
+  - multiple sequence alignment
+  - quality control
+tools:
+  - "goalign":
+      description: "goalign is a set of command line tools to manipulate multiple alignments"
+      homepage: "https://github.com/evolbioinfo/goalign"
+      documentation: "https://github.com/evolbioinfo/goalign/blob/master/docs/index.md"
+      tool_dev_url: "https://github.com/evolbioinfo/goalign"
+      doi: "10.1093/nargab/lqab075"
+      licence: "['GPL v2']"
+
+input:
+  - aln:
+      type: file
+      description: MultiFASTA/Phylip alignment file
+      pattern: "*.{fasta,phy}"
+
+output:
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - aln:
+      type: file
+      description: MultiFASTA/Phylip alignment file
+      pattern: "*.{fasta,phy}"
-      pattern: "*.{fasta,phy}"
+      pattern: "*.{fasta,fas,fa,phy}"
-      pattern: "*.{fasta,phy}"
+      pattern: "*.{fasta,fas,fa,phy}"
+
+authors:
+  - "@andersgs"
@@ -1123,6 +1123,10 @@ goat/taxonsearch:
   - modules/nf-core/goat/taxonsearch/**
   - tests/modules/nf-core/goat/taxonsearch/**
 
+goalign/cleanseqs:
+  - modules/goalign/cleanseqs/**
+  - tests/modules/goalign/cleanseqs/**
+
 graphmap2/align:
   - modules/nf-core/graphmap2/align/**
   - tests/modules/nf-core/graphmap2/align/**

diff --git a/tests/modules/goalign/cleanseqs/main.nf b/tests/modules/goalign/cleanseqs/main.nf
@@ -0,0 +1,12 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GOALIGN_CLEANSEQS } from '../../../../modules/goalign/cleanseqs/main.nf'
+
+workflow test_goalign_cleanseqs {
+
+    input = file(params.test_data['sarscov2']['genome']['all_sites_fas'], checkIfExists: true)
+
+    GOALIGN_CLEANSEQS ( input )
+}
diff --git a/tests/modules/goalign/cleanseqs/nextflow.config b/tests/modules/goalign/cleanseqs/nextflow.config
@@ -0,0 +1,5 @@
+process {
+    ext.args = "--cutoff 0.05"
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}
diff --git a/tests/modules/goalign/cleanseqs/test.yml b/tests/modules/goalign/cleanseqs/test.yml
@@ -0,0 +1,12 @@
+- name: "goalign cleanseqs"
+  command: nextflow run ./tests/modules/goalign/cleanseqs -entry test_goalign_cleanseqs -c ./tests/config/nextflow.config -c ./tests/modules/goalign/cleanseqs/nextflow.config
+  tags:
+    - "goalign"
+    #
+    - "goalign/cleanseqs"
+    #
+  files:
+    - path: "output/goalign/all_sites.fas_clean.fas"
+      md5sum: f7b946d6d5000f83baa401addd7bb26a
+    - path: output/goalign/versions.yml
+      md5sum: 2dc17f26bc12547017e2a7eb66ae503f