Merge pull request #172 from andersgs/ivar-variant

Add ivar variants

.github/filters.yml

@@ -148,6 +148,10 @@ ivar_trim:
   - software/ivar/trim/**
   - tests/software/ivar/trim/**
 
+ivar_variants:
+  - software/ivar/variants/**
+  - tests/software/ivar/variants/**
+
 minimap2_align:
   - software/minimap2/align/**
   - tests/software/minimap2/align/**

software/ivar/variants/functions.nf

@@ -0,0 +1,59 @@
+/*
+ * -----------------------------------------------------
+ *  Utility functions used in nf-core DSL2 module files
+ * -----------------------------------------------------
+ */
+
+/*
+ * Extract name of software tool from process name using $task.process
+ */
+def getSoftwareName(task_process) {
+    return task_process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()
+}
+
+/*
+ * Function to initialise default values and to generate a Groovy Map of available options for nf-core modules
+ */
+def initOptions(Map args) {
+    def Map options = [:]
+    options.args          = args.args ?: ''
+    options.args2         = args.args2 ?: ''
+    options.publish_by_id = args.publish_by_id ?: false
+    options.publish_dir   = args.publish_dir ?: ''
+    options.publish_files = args.publish_files
+    options.suffix        = args.suffix ?: ''
+    return options
+}
+
+/*
+ * Tidy up and join elements of a list to return a path string
+ */
+def getPathFromList(path_list) {
+    def paths = path_list.findAll { item -> !item?.trim().isEmpty() }  // Remove empty entries
+    paths = paths.collect { it.trim().replaceAll("^[/]+|[/]+\$", "") } // Trim whitespace and trailing slashes
+    return paths.join('/')
+}
+
+/*
+ * Function to save/publish module results
+ */
+def saveFiles(Map args) {
+    if (!args.filename.endsWith('.version.txt')) {
+        def ioptions = initOptions(args.options)
+        def path_list = [ ioptions.publish_dir ?: args.publish_dir ]
+        if (ioptions.publish_by_id) {
+            path_list.add(args.publish_id)
+        }
+        if (ioptions.publish_files instanceof Map) {
+            for (ext in ioptions.publish_files) {
+                if (args.filename.endsWith(ext.key)) {
+                    def ext_list = path_list.collect()
+                    ext_list.add(ext.value)
+                    return "${getPathFromList(ext_list)}/$args.filename"
+                }
+            }
+        } else if (ioptions.publish_files == null) {
+            return "${getPathFromList(path_list)}/$args.filename"
+        }
+    }
+}

software/ivar/variants/main.nf

@@ -0,0 +1,50 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+
+params.options = [:]
+def options    = initOptions(params.options)
+
+process IVAR_VARIANTS {
+    tag "$meta.id"
+    label 'process_medium'
+    publishDir "${params.outdir}",
+        mode: params.publish_dir_mode,
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), publish_id:meta.id) }
+
+    conda (params.enable_conda ? "bioconda::ivar=1.3.1" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/ivar:1.3.1--h089eab3_0"
+    } else {
+        container "quay.io/biocontainers/ivar:1.3.1--h089eab3_0"
+    }
+
+    input:
+    tuple val(meta), path(bam)
+    path  fasta
+    path  gff
+
+    output:
+    tuple val(meta), path("*.tsv")    , emit: tsv
+    tuple val(meta), path("*.mpileup"), optional:true, emit: mpileup
+    path "*.version.txt"              , emit: version
+
+    script:
+    def software     = getSoftwareName(task.process)
+    def prefix       = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    def save_mpileup = params.save_mpileup ? "tee ${prefix}.mpileup |" : ""
+    def features     = params.gff ? "-g $gff" : ""
+    """
+    samtools mpileup \\
+        $options.args2 \\
+        --reference $fasta \\
+        $bam | \\
+        $save_mpileup  \\
+        ivar variants \\
+            $options.args \\
+            $features \\
+            -r $fasta \\
+            -p $prefix
+
+    ivar version | head -n1 2>&1 | sed 's/^.*iVar version //g' > ${software}.version.txt
+    """
+}

software/ivar/variants/meta.yml

@@ -0,0 +1,72 @@
+name: ivar_variants
+description: Call variants from a BAM file using iVar
+keywords:
+    - amplicon sequencing
+    - variants
+    - fasta
+tools:
+    - ivar:
+        description: |
+            iVar - a computational package that contains functions broadly useful for viral amplicon-based sequencing.
+        homepage: https://github.com/andersen-lab/ivar
+        documentation: https://andersen-lab.github.io/ivar/html/manualpage.html
+params:
+    - outdir:
+        type: string
+        description: |
+            The pipeline's output directory. By default, the module will
+            output files into `$params.outdir/<SOFTWARE>`
+    - publish_dir_mode:
+        type: string
+        description: |
+            Value for the Nextflow `publishDir` mode parameter.
+            Available: symlink, rellink, link, copy, copyNoFollow, move.
+    - enable_conda:
+        type: boolean
+        description: |
+            Run the module with Conda using the software specified
+            via the `conda` directive
+    - singularity_pull_docker_container:
+        type: boolean
+        description: |
+            Instead of directly downloading Singularity images for use with Singularity,
+            force the workflow to pull and convert Docker containers instead.
+input:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - bam:
+        type: file
+        description: A sorted (with samtools sort) and trimmed (with iVar trim) bam file
+        pattern: "*.bam"
+    - fasta:
+        type: file
+        description: The reference sequence used for mapping and generating the BAM file
+        pattern: "*.fa"
+    - gff:
+        type: file
+        description: A GFF file in the GFF3 format can be supplied to specify coordinates of open reading frames (ORFs). In absence of GFF file, amino acid translation will not be done.
+        patter: "*.gff"
+output:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - tsv:
+        type: file
+        description: iVar generated TSV file with the variants
+        pattern: "*.tsv"
+    - mpileup:
+        type: file
+        description: mpileup output from samtools mpileup [OPTIONAL]
+        pattern: "*.mpileup"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@andersgs"
+    - "@drpatelh"

tests/data/gff/sarscov2/MN908947.3.gff3

@@ -0,0 +1,27 @@
+##sequence-region MN908947.3 1 29903
+##species https://www.ncbi.nlm.nih.gov/Taxonomy/Browser/wwwtax.cgi?id=2697049
+MN908947.3	Genbank	region	1	29903	.	+	.	ID=MN908947.3:1..29903;Dbxref=taxon:2697049;collection-date=Dec-2019;country=China;gb-acronym=SARS-CoV-2;gbkey=Src;genome=genomic;isolate=Wuhan-Hu-1;mol_type=genomic RNA;nat-host=Homo sapiens;old-name=Wuhan seafood market pneumonia virus
+MN908947.3	Genbank	five_prime_UTR	1	265	.	+	.	ID=id-MN908947.3:1..265;gbkey=5'UTR
+MN908947.3	Genbank	gene	266	21555	.	+	.	ID=gene-orf1ab;Name=orf1ab;gbkey=Gene;gene=orf1ab;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	266	13468	.	+	0	ID=cds-QHD43415.1;Parent=gene-orf1ab;Dbxref=NCBI_GP:QHD43415.1;Name=QHD43415.1;Note=translated by -1 ribosomal frameshift;exception=ribosomal slippage;gbkey=CDS;gene=orf1ab;part=1;product=orf1ab polyprotein;protein_id=QHD43415.1
+MN908947.3	Genbank	CDS	13468	21555	.	+	0	ID=cds-QHD43415.1;Parent=gene-orf1ab;Dbxref=NCBI_GP:QHD43415.1;Name=QHD43415.1;Note=translated by -1 ribosomal frameshift;exception=ribosomal slippage;gbkey=CDS;gene=orf1ab;part=2;product=orf1ab polyprotein;protein_id=QHD43415.1
+MN908947.3	Genbank	gene	21563	25384	.	+	.	ID=gene-S;Name=S;gbkey=Gene;gene=S;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	21563	25384	.	+	0	ID=cds-QHD43416.1;Parent=gene-S;Dbxref=NCBI_GP:QHD43416.1;Name=QHD43416.1;Note=structural protein;gbkey=CDS;gene=S;product=surface glycoprotein;protein_id=QHD43416.1
+MN908947.3	Genbank	gene	25393	26220	.	+	.	ID=gene-ORF3a;Name=ORF3a;gbkey=Gene;gene=ORF3a;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	25393	26220	.	+	0	ID=cds-QHD43417.1;Parent=gene-ORF3a;Dbxref=NCBI_GP:QHD43417.1;Name=QHD43417.1;gbkey=CDS;gene=ORF3a;product=ORF3a protein;protein_id=QHD43417.1
+MN908947.3	Genbank	gene	26245	26472	.	+	.	ID=gene-E;Name=E;gbkey=Gene;gene=E;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	26245	26472	.	+	0	ID=cds-QHD43418.1;Parent=gene-E;Dbxref=NCBI_GP:QHD43418.1;Name=QHD43418.1;Note=structural protein%3B E protein;gbkey=CDS;gene=E;product=envelope protein;protein_id=QHD43418.1
+MN908947.3	Genbank	gene	26523	27191	.	+	.	ID=gene-M;Name=M;gbkey=Gene;gene=M;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	26523	27191	.	+	0	ID=cds-QHD43419.1;Parent=gene-M;Dbxref=NCBI_GP:QHD43419.1;Name=QHD43419.1;Note=structural protein;gbkey=CDS;gene=M;product=membrane glycoprotein;protein_id=QHD43419.1
+MN908947.3	Genbank	gene	27202	27387	.	+	.	ID=gene-ORF6;Name=ORF6;gbkey=Gene;gene=ORF6;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	27202	27387	.	+	0	ID=cds-QHD43420.1;Parent=gene-ORF6;Dbxref=NCBI_GP:QHD43420.1;Name=QHD43420.1;gbkey=CDS;gene=ORF6;product=ORF6 protein;protein_id=QHD43420.1
+MN908947.3	Genbank	gene	27394	27759	.	+	.	ID=gene-ORF7a;Name=ORF7a;gbkey=Gene;gene=ORF7a;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	27394	27759	.	+	0	ID=cds-QHD43421.1;Parent=gene-ORF7a;Dbxref=NCBI_GP:QHD43421.1;Name=QHD43421.1;gbkey=CDS;gene=ORF7a;product=ORF7a protein;protein_id=QHD43421.1
+MN908947.3	Genbank	gene	27894	28259	.	+	.	ID=gene-ORF8;Name=ORF8;gbkey=Gene;gene=ORF8;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	27894	28259	.	+	0	ID=cds-QHD43422.1;Parent=gene-ORF8;Dbxref=NCBI_GP:QHD43422.1;Name=QHD43422.1;gbkey=CDS;gene=ORF8;product=ORF8 protein;protein_id=QHD43422.1
+MN908947.3	Genbank	gene	28274	29533	.	+	.	ID=gene-N;Name=N;gbkey=Gene;gene=N;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	28274	29533	.	+	0	ID=cds-QHD43423.2;Parent=gene-N;Dbxref=NCBI_GP:QHD43423.2;Name=QHD43423.2;Note=structural protein;gbkey=CDS;gene=N;product=nucleocapsid phosphoprotein;protein_id=QHD43423.2
+MN908947.3	Genbank	gene	29558	29674	.	+	.	ID=gene-ORF10;Name=ORF10;gbkey=Gene;gene=ORF10;gene_biotype=protein_coding
+MN908947.3	Genbank	CDS	29558	29674	.	+	0	ID=cds-QHI42199.1;Parent=gene-ORF10;Dbxref=NCBI_GP:QHI42199.1;Name=QHI42199.1;gbkey=CDS;gene=ORF10;product=ORF10 protein;protein_id=QHI42199.1
+MN908947.3	Genbank	three_prime_UTR	29675	29903	.	+	.	ID=id-MN908947.3:29675..29903;gbkey=3'UTR
+

tests/software/ivar/variants/main.nf

@@ -0,0 +1,41 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { IVAR_VARIANTS } from '../../../../software/ivar/variants/main.nf' addParams([:])
+
+workflow test_ivar_variants_no_gff_no_mpileup {
+    params.gff          = false
+    params.save_mpileup = false
+    
+    def ref   = file("${launchDir}/tests/data/fasta/sarscov2/MN908947.3.fa", checkIfExists: true)
+    def dummy = file("${launchDir}/tests/data/dummy/dummy_file.txt", checkIfExists: true)
+    def input = []
+    input = [ [ id:'test'], 
+              file("${launchDir}/tests/data/bam/test-sc2-artic-v3-sorted-trimmed.bam", checkIfExists: true) ]
+    IVAR_VARIANTS ( input, ref, dummy )
+}
+
+workflow test_ivar_variants_no_gff_with_mpileup {
+    params.gff          = false
+    params.save_mpileup = true
+    
+    def ref   = file("${launchDir}/tests/data/fasta/sarscov2/MN908947.3.fa", checkIfExists: true)
+    def dummy = file("${launchDir}/tests/data/dummy/dummy_file.txt", checkIfExists: true)
+    def input = []
+    input = [ [ id:'test'], 
+              file("${launchDir}/tests/data/bam/test-sc2-artic-v3-sorted-trimmed.bam", checkIfExists: true) ]
+    IVAR_VARIANTS ( input, ref, dummy )
+}
+
+workflow test_ivar_variants_with_gff_with_mpileup {
+    params.gff          = "${launchDir}/tests/data/gff/sarscov2/MN908947.3.gff3"
+    params.save_mpileup = true
+
+    def ref = file("${launchDir}/tests/data/fasta/sarscov2/MN908947.3.fa", checkIfExists: true)
+    def gff = file(params.gff, checkIfExists: true)
+    def input = []
+    input = [ [ id:'test'], 
+              file("${launchDir}/tests/data/bam/test-sc2-artic-v3-sorted-trimmed.bam", checkIfExists: true) ]
+    IVAR_VARIANTS ( input, ref, gff )
+}

tests/software/ivar/variants/test.yml

@@ -0,0 +1,30 @@
+- name: ivar variants no gff no mpileup
+  command: nextflow run ./tests/software/ivar/variants -entry test_ivar_variants_no_gff_no_mpileup -c tests/config/nextflow.config
+  tags:
+    - ivar
+    - ivar_variants
+  files:
+    - path: output/ivar/test.tsv
+      md5sum: 3fd376b8ad80f27a999edf3345d2014a
+      
+- name: ivar variants no gff with mpileup
+  command: nextflow run ./tests/software/ivar/variants -entry test_ivar_variants_no_gff_with_mpileup -c tests/config/nextflow.config --save_mpileup
+  tags:
+    - ivar
+    - ivar_variants
+  files:
+    - path: output/ivar/test.tsv
+      md5sum: 3fd376b8ad80f27a999edf3345d2014a
+    - path: output/ivar/test.mpileup
+      md5sum: b3dfd337dfcd313d6cc9680d7e796fe6
+
+- name: ivar variants with gff with mpileup
+  command: nextflow run ./tests/software/ivar/variants -entry test_ivar_variants_with_gff_with_mpileup -c tests/config/nextflow.config --gff tests/data/gff/sarscov2/MN908947.3.gff3 --save_mpileup
+  tags:
+    - ivar
+    - ivar_variants
+  files:
+    - path: output/ivar/test.tsv
+      md5sum: 3fd376b8ad80f27a999edf3345d2014a
+    - path: output/ivar/test.mpileup
+      md5sum: b3dfd337dfcd313d6cc9680d7e796fe6