feat: add sub-workflow for SRA (#836)

* feat: add sub-workflow for SRA
* Combine prefetch and fasterq-dump into one sub-workflow
* tests: add sub-workflow to pytest config

subworkflows/nf-core/sra_fastq/main.nf

@@ -0,0 +1,34 @@
+//
+// Download FASTQ sequencing reads from the NCBI's Sequence Read Archive (SRA).
+//
+
+params.prefetch_options    = [:]
+params.fasterqdump_options = [:]
+
+include { SRATOOLS_PREFETCH    } from '../../../modules/sratools/prefetch/main'    addParams( options: params.prefetch_options    )
+include { SRATOOLS_FASTERQDUMP } from '../../../modules/sratools/fasterqdump/main' addParams( options: params.fasterqdump_options )
+
+workflow SRA_FASTQ {
+    take:
+    sra_ids  // channel: [ val(meta), val(id) ]
+
+    main:
+
+    ch_versions = Channel.empty()
+
+    //
+    // Prefetch sequencing reads in SRA format.
+    //
+    SRATOOLS_PREFETCH ( sra_ids )
+    ch_versions = ch_versions.mix( SRATOOLS_PREFETCH.out.versions.first() )
+
+    //
+    // Convert the SRA format into one or more compressed FASTQ files.
+    //
+    SRATOOLS_FASTERQDUMP ( SRATOOLS_PREFETCH.out.sra )
+    ch_versions = ch_versions.mix( SRATOOLS_FASTERQDUMP.out.versions.first() )
+
+    emit:
+    reads    = SRATOOLS_FASTERQDUMP.out.reads  // channel: [ val(meta), [ reads ] ]
+    versions = ch_versions                     // channel: [ versions.yml ]
+}

subworkflows/nf-core/sra_fastq/meta.yml

@@ -0,0 +1,37 @@
+name: sra_fastq
+description: Download FASTQ sequencing reads from the NCBI's Sequence Read Archive (SRA).
+keywords:
+  - sequencing
+  - FASTQ
+  - prefetch
+  - dump
+modules:
+  - sratools/prefetch
+  - sratools/fasterqdump
+input:
+  - meta:
+      type: map
+      description: >
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - id:
+      type: string
+      description: >
+        SRA identifier.
+# TODO Update when we decide on a standard for subworkflow docs
+output:
+  - meta:
+      type: map
+      description: >
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - reads:
+      type: file
+      description: Extracted FASTQ file or files if the sequencing reads are paired-end.
+      pattern: "*.fastq.gz"
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+authors:
+  - '@Midnighter'

subworkflows/nf-core/sra_fastq/nextflow.config

@@ -0,0 +1,2 @@
+params.prefetch_options    = [:]
+params.fasterqdump_options = [:]

tests/config/pytest_subworkflows.yml

@@ -9,3 +9,8 @@ subworkflows/bam_stats_samtools:
 subworkflows/bam_sort_samtools:
   - subworkflows/nf-core/bam_sort_samtools/**
   - tests/subworkflows/nf-core/bam_sort_samtools/**
+
+subworkflows/sra_fastq:
+  - subworkflows/nf-core/sra_fastq/**
+  - tests/subworkflows/nf-core/sra_fastq/**
+

tests/subworkflows/nf-core/sra_fastq/main.nf

@@ -0,0 +1,23 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SRA_FASTQ } from '../../../../subworkflows/nf-core/sra_fastq/main.nf' addParams( [:] )
+
+workflow test_sra_fastq_single_end {
+    input = [
+        [ id:'test_single_end', single_end:true ], // meta map
+        'SRR13255544'
+    ]
+
+    SRA_FASTQ ( input )
+}
+
+workflow test_sra_fastq_paired_end {
+    input = [
+        [ id:'test_paired_end', single_end:false ], // meta map
+        'SRR11140744'
+    ]
+
+    SRA_FASTQ ( input )
+}

tests/subworkflows/nf-core/sra_fastq/test.yml

@@ -0,0 +1,25 @@
+- name: sra fastq single-end
+  command: nextflow run ./tests/subworkflows/nf-core/sra_fastq -entry test_sra_fastq_single_end -c tests/config/nextflow.config
+  tags:
+    - subworkflows/sra_fastq
+    # Modules
+    - sratools
+    - sratools/prefetch
+    - sratools/fasterqdump
+  files:
+    - path: output/sratools/SRR13255544.fastq.gz
+      md5sum: 1054c7b71884acdb5eed8a378f18be82
+
+- name: sra fastq paired-end
+  command: nextflow run ./tests/subworkflows/nf-core/sra_fastq -entry test_sra_fastq_paired_end -c tests/config/nextflow.config
+  tags:
+    - subworkflows/sra_fastq
+    # Modules
+    - sratools
+    - sratools/prefetch
+    - sratools/fasterqdump
+  files:
+    - path: output/sratools/SRR11140744_1.fastq.gz
+      md5sum: 193809c784a4ea132ab2a253fa4f55b6
+    - path: output/sratools/SRR11140744_2.fastq.gz
+      md5sum: 3e3b3af3413f50a1685fd7b3f1456d4e