Merge pull request #221 from heuermh/seqwish

Add module for seqwish/induce

.github/filters.yml

@@ -229,6 +229,10 @@ seacr_callpeak:
   - software/seacr/callpeak/**
   - tests/software/seacr/callpeak/**
 
+seqwish_induce:
+  - software/seqwish/induce/**
+  - tests/software/seqwish/induce/**
+
 star_align:
   - software/star/align/**
   - tests/software/star/align/**

software/seqwish/induce/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/seqwish/induce/main.nf

@@ -0,0 +1,44 @@
+// Import generic module functions
+include { initOptions; saveFiles; getSoftwareName } from './functions'
+params.options = [:]
+def options    = initOptions(params.options)
+
+def VERSION = '0.4.1'
+
+process SEQWISH_INDUCE {
+    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::seqwish=0.4.1" : null)
+
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/seqwish:0.4.1--h8b12597_0"
+    } else {
+        container "quay.io/biocontainers/seqwish:0.4.1--h8b12597_0"
+    }
+
+    input:
+    tuple val(meta), path(paf), path(fasta)
+
+    output:
+    tuple val(meta), path("*.gfa"), emit: gfa
+    path "*.version.txt"          , emit: version
+
+
+    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
+    """
+    seqwish \\
+        --threads $task.cpus \\
+        --paf-alns=$paf \\
+        --seqs=$fasta \\
+        --gfa=${prefix}.gfa \\
+        $options.args
+
+    echo $VERSION > ${software}.version.txt
+    """
+}

software/seqwish/induce/meta.yml

@@ -0,0 +1,67 @@
+name: seqwish_induce
+description: Induce a variation graph in GFA format from alignments in PAF format
+keywords:
+    - induce
+    - paf
+    - gfa
+    - graph
+    - variation graph
+tools:
+    - seqwish:
+        description: |
+            seqwish implements a lossless conversion from pairwise alignments between
+            sequences to a variation graph encoding the sequences and their alignments.
+        homepage: https://github.com/ekg/seqwish
+        documentation: https://github.com/ekg/seqwish
+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 ]
+    - paf:
+        type: file
+        description: PAF file of alignments
+        pattern: "*.{paf,paf.gz}"
+    - fasta:
+        type: file
+        description: FASTA file used to generate alignments
+        pattern: "*.{fa,fa.gz,fasta,fasta.gz}"
+
+output:
+    - meta:
+        type: map
+        description: |
+            Groovy Map containing sample information
+            e.g. [ id:'test', single_end:false ]
+    - gfa:
+        type: file
+        description: Variation graph in GFA1 format
+        pattern: "*.{gfa}"
+    - version:
+        type: file
+        description: File containing software version
+        pattern: "*.{version.txt}"
+authors:
+    - "@heuermh"

tests/data/paf/GCA_011545545.1_ASM1154554v1_cds_from_genomic.paf

@@ -0,0 +1,11 @@
+lcl|MT192765.1_cds_QIK50426.1_1	21291	1	21289	+	lcl|MT192765.1_cds_QIK50426.1_1	21291	1	21289	21288	21288	60	tp:A:P	cm:i:3984	s1:i:21288	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50427.1_2	3822	8	3821	+	lcl|MT192765.1_cds_QIK50427.1_2	3822	8	3821	3813	3813	60	tp:A:P	cm:i:722	s1:i:3813	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50428.1_3	828	4	819	+	lcl|MT192765.1_cds_QIK50428.1_3	828	4	819	815	815	60	tp:A:P	cm:i:148	s1:i:815	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50429.1_4	228	6	226	+	lcl|MT192765.1_cds_QIK50429.1_4	228	6	226	220	220	60	tp:A:P	cm:i:42	s1:i:220	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50430.1_5	669	6	663	+	lcl|MT192765.1_cds_QIK50430.1_5	669	6	663	657	657	60	tp:A:P	cm:i:115	s1:i:657	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50431.1_6	186	0	185	+	lcl|MT192765.1_cds_QIK50431.1_6	186	0	185	185	185	60	tp:A:P	cm:i:33	s1:i:185	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50432.1_7	366	1	358	+	lcl|MT192765.1_cds_QIK50432.1_7	366	1	358	357	357	60	tp:A:P	cm:i:64	s1:i:357	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50433.1_8	132	6	124	+	lcl|MT192765.1_cds_QIK50433.1_8	132	6	124	118	118	60	tp:A:P	cm:i:20	s1:i:118	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50434.1_9	366	6	357	+	lcl|MT192765.1_cds_QIK50434.1_9	366	6	357	351	351	60	tp:A:P	cm:i:68	s1:i:351	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50435.1_10	1260	4	1255	+	lcl|MT192765.1_cds_QIK50435.1_10	1260	4	1255	1251	1251	60	tp:A:P	cm:i:231	s1:i:1251	s2:i:0	dv:f:0	rl:i:0
+lcl|MT192765.1_cds_QIK50436.1_11	117	0	116	+	lcl|MT192765.1_cds_QIK50436.1_11	117	0	116	116	116	60	tp:A:P	cm:i:19	s1:i:116	s2:i:0	dv:f:0	rl:i:0

tests/software/seqwish/induce/main.nf

@@ -0,0 +1,15 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SEQWISH_INDUCE } from '../../../../software/seqwish/induce/main.nf' addParams( options: [:] )
+
+workflow test_seqwish_induce {
+
+    def input = []
+    input = [ [ id:'GCA_011545545.1_ASM1154554v1_cds_from_genomic' ], // meta map
+              [ file("${launchDir}/tests/data/paf/GCA_011545545.1_ASM1154554v1_cds_from_genomic.paf", checkIfExists: true) ],
+              [ file("${launchDir}/tests/data/fasta/sarscov2/GCA_011545545.1_ASM1154554v1_cds_from_genomic.fna", checkIfExists: true) ] ]
+
+    SEQWISH_INDUCE ( input )
+}

tests/software/seqwish/induce/test.yml

@@ -0,0 +1,8 @@
+- name: seqwish induce
+  command: nextflow run ./tests/software/seqwish/induce -entry test_seqwish_induce -c tests/config/nextflow.config
+  tags:
+    - seqwish
+    - seqwish_induce
+  files:
+    - path: output/seqwish/GCA_011545545.1_ASM1154554v1_cds_from_genomic.gfa
+      md5sum: 216a02d3aca322a457c31a62c628548e