add snippy/core module (#1855)

* add snippy-core module

* Update modules/snippy/core/meta.yml

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* Update tests/modules/snippy/core/main.nf

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* Update tests/modules/snippy/core/main.nf

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* Update modules/snippy/core/main.nf

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* Update main.nf

* update test

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

modules/snippy/core/main.nf

@@ -0,0 +1,53 @@
+process SNIPPY_CORE {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::snippy=4.6.0" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/snippy:4.6.0--hdfd78af_2':
+        'quay.io/biocontainers/snippy:4.6.0--hdfd78af_1' }"
+
+    input:
+    tuple val(meta), path(vcf), path(aligned_fa)
+    path reference
+
+    output:
+    tuple val(meta), path("${prefix}.aln")     , emit: aln
+    tuple val(meta), path("${prefix}.full.aln"), emit: full_aln
+    tuple val(meta), path("${prefix}.tab")     , emit: tab
+    tuple val(meta), path("${prefix}.vcf")     , emit: vcf
+    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}"
+    def is_compressed = reference.getName().endsWith(".gz") ? true : false
+    def reference_name = reference.getName().replace(".gz", "")
+    """
+    if [ "$is_compressed" == "true" ]; then
+        gzip -c -d $reference > $reference_name
+    fi
+
+    # Collect samples into necessary folders
+    mkdir samples
+    find . -name "*.vcf" | sed 's/.vcf//' | xargs -I {} bash -c 'mkdir samples/{}'
+    find . -name "*.vcf" | sed 's/.vcf//' | xargs -I {} bash -c 'cp -L {}.vcf samples/{}/{}.vcf'
+    find . -name "*.aligned.fa" | sed 's/.aligned.fa//' | xargs -I {} bash -c 'cp -L {}.aligned.fa samples/{}/{}.aligned.fa'
+
+    # Run snippy-core
+    snippy-core \\
+        $args \\
+        --ref $reference_name \\
+        --prefix $prefix \\
+        samples/*
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        snippy-core: \$(echo \$(snippy-core --version 2>&1) | sed 's/snippy-core //')
+    END_VERSIONS
+    """
+}

modules/snippy/core/meta.yml

@@ -0,0 +1,67 @@
+name: snippy_core
+description: Core-SNP alignment from Snippy outputs
+keywords:
+  - core
+  - alignment
+  - bacteria
+  - snippy
+tools:
+  - snippy:
+      description: "Rapid bacterial SNP calling and core genome alignments"
+      homepage: "https://github.com/tseemann/snippy"
+      documentation: "https://github.com/tseemann/snippy"
+      tool_dev_url: "https://github.com/tseemann/snippy"
+      doi: ""
+      licence: "['GPL v2']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - vcf:
+      type: file
+      description: Annotated variants in VCF format
+      pattern: "*.vcf.gz"
+  - aligned_fa:
+      type: file
+      description: A version of the reference but with - at position with depth=0 and N for 0 < depth < --mincov (does not have variants)
+      pattern: "*.aligned.fa.gz"
+  - reference:
+      type: file
+      description: Reference genome in GenBank (preferred) or FASTA format
+      pattern: "*.{gbk,gbk.gz,gbff,gbff.gz,fa,fa.gz}"
+
+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"
+  - aln:
+      type: file
+      description: A core SNP alignment in FASTA format
+      pattern: "*.aln"
+  - full_aln:
+      type: file
+      description: A whole genome SNP alignment (includes invariant sites)
+      pattern: "*.full.aln"
+  - tab:
+      type: file
+      description: Tab-separated columnar list of core SNP sites with alleles but NO annotations
+      pattern: "*.tab"
+  - vcf:
+      type: file
+      description: Multi-sample VCF file with genotype GT tags for all discovered alleles
+      pattern: "*.vcf"
+  - txt:
+      type: file
+      description: Tab-separated columnar list of alignment/core-size statistics
+      pattern: "*.txt"
+authors:
+  - "@rpetit3"

tests/config/pytest_modules.yml

@@ -2159,6 +2159,10 @@ snapaligner/index:
   - modules/snapaligner/index/**
   - tests/modules/snapaligner/index/**
 
+snippy/core:
+  - modules/snippy/core/**
+  - tests/modules/snippy/core/**
+
 snippy/run:
   - modules/snippy/run/**
   - tests/modules/snippy/run/**

tests/modules/snippy/core/main.nf

@@ -0,0 +1,22 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SNIPPY_RUN } from '../../../../modules/snippy/run/main.nf'
+include { SNIPPY_CORE } from '../../../../modules/snippy/core/main.nf'
+
+workflow test_snippy_core {
+    
+    input = [ [ id:'test', single_end:false ], // meta map
+              [ file(params.test_data['candidatus_portiera_aleyrodidarum']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+                file(params.test_data['candidatus_portiera_aleyrodidarum']['illumina']['test_2_fastq_gz'], checkIfExists: true) ]
+            ]
+    reference = file(params.test_data['candidatus_portiera_aleyrodidarum']['genome']['genome_fasta'], checkIfExists: true)
+
+    SNIPPY_RUN ( input, reference )
+    SNIPPY_RUN.out.vcf.collect{meta, vcf -> vcf}.map{ vcf -> [[id:'snippy-core'], vcf]}.set{ ch_merge_vcf }
+    SNIPPY_RUN.out.aligned_fa.collect{meta, aligned_fa -> aligned_fa}.map{ aligned_fa -> [[id:'snippy-core'], aligned_fa]}.set{ ch_merge_aligned_fa }
+    ch_merge_vcf.join( ch_merge_aligned_fa ).set{ ch_snippy_core }
+    SNIPPY_CORE( ch_snippy_core, reference )
+
+}

tests/modules/snippy/core/nextflow.config

@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    
+}

tests/modules/snippy/core/test.yml

@@ -0,0 +1,16 @@
+- name: snippy core test_snippy_core
+  command: nextflow run ./tests/modules/snippy/core -entry test_snippy_core -c ./tests/config/nextflow.config  -c ./tests/modules/snippy/core/nextflow.config
+  tags:
+    - snippy
+    - snippy/core
+  files:
+    - path: output/snippy/snippy-core.aln
+      md5sum: 23cf80454a83849977aeb7193a5eab8e
+    - path: output/snippy/snippy-core.full.aln
+      md5sum: fcadf17c2c8109fe4491ab7e7df6e261
+    - path: output/snippy/snippy-core.tab
+      md5sum: 1eb437b4d82d8dbc93649b0edf94abc8
+    - path: output/snippy/snippy-core.txt
+      md5sum: 243961814f44b9a6cac48b69f02b4349
+    - path: output/snippy/snippy-core.vcf
+      md5sum: 8b4ff1f2e7f1cef9d168dad5d70e642a