Fixes for nanoseq modules (#479)

* Fix minimap2 index module * Fix minimap2 index tests * Fix graphmap2 index module * Fix graphmap2 module * Fix ECLint * Fix bedtools bamtobed module * Fix tests for bedtools bamtobed module * Add tag for graphmap2 align module * Fix EClint * Fix qcat module * Add md5sum for graphmap2/align module * Remove non-started test data file * Remove md5sum for graphmap2 align
2024-12-31 03:32:10 -05:00 · 2021-04-30 15:57:43 +01:00 · 2021-04-30 15:57:43 +01:00 · 466ab67808
commit 466ab67808
parent 05f479f03a
16 changed files with 115 additions and 80 deletions
--- a/software/bedtools/bamtobed/main.nf
+++ b/software/bedtools/bamtobed/main.nf
@ -2,7 +2,7 @@
 include { initOptions; saveFiles; getSoftwareName } from './functions'

 params.options = [:]
-def options    = initOptions(params.options)
+options        = initOptions(params.options)

 process BEDTOOLS_BAMTOBED {
    tag "$meta.id"
@ -11,24 +11,30 @@ process BEDTOOLS_BAMTOBED {
        mode: params.publish_dir_mode,
        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['id']) }

-    conda     (params.enable_conda ? "bioconda::bedtools=2.29.2" : null)
-    container "quay.io/biocontainers/bedtools:2.29.2--hc088bd4_0"
+    conda (params.enable_conda ? "bioconda::bedtools=2.30.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/bedtools:2.30.0--hc088bd4_0"
+    } else {
+        container "quay.io/biocontainers/bedtools:2.30.0--hc088bd4_0"
+    }

    input:
-    tuple val(meta), path(sizes), path(bam), path(bai)
+    tuple val(meta), path(bam)

    output:
-    tuple val(meta), path(sizes), path("*.bed12"), emit: bed12
-    path "*.version.txt"                         , emit: version
+    tuple val(meta), path("*.bed"), emit: bed
+    path  "*.version.txt"         , emit: version

    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
    """
    bedtools \\
        bamtobed \\
-        -bed12 \\
-        -cigar \\
-        -i ${bam[0]} \\
-        | bedtools sort > ${meta.id}.bed12
-    bedtools --version | sed -e "s/bedtools v//g" > bedtools.version.txt
+        $options.args \\
+        -i $bam \\
+        | bedtools sort > ${prefix}.bed
+
+    bedtools --version | sed -e "s/bedtools v//g" > ${software}.version.txt
    """
 }
--- a/software/bedtools/bamtobed/meta.yml
+++ b/software/bedtools/bamtobed/meta.yml
@ -18,19 +18,15 @@ input:
        type: file
        description: Input BAM file
        pattern: "*.{bam}"
-    - sizes:
-        type: file
-        description: File which defines the chromosome lengths for a given genome
-        pattern: "*.{sizes}"
 output:
    - meta:
        type: map
        description: |
            Groovy Map containing sample information
            e.g. [ id:'test', single_end:false ]
-    - bed12:
+    - bed:
        type: file
-        description: Bed file with 12 columns containing all genomic intervals.
+        description: Bed file containing genomic intervals.
        pattern: "*.{bed}"
    - version:
        type: file
@ -38,3 +34,4 @@ output:
        pattern: "*.{version.txt}"
 authors:
    - "@yuukiiwa"
+    - "@drpatelh"
--- a/software/graphmap2/align/main.nf
+++ b/software/graphmap2/align/main.nf
@ -2,34 +2,45 @@
 include { initOptions; saveFiles; getSoftwareName } from './functions'

 params.options = [:]
-def options    = initOptions(params.options)
+options        = initOptions(params.options)

 process GRAPHMAP2_ALIGN {
+    tag "$meta.id"
+    label 'process_medium'
    tag "$meta.id"
    publishDir "${params.outdir}",
        mode: params.publish_dir_mode,
        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['id']) }

-    conda     (params.enable_conda ? "bioconda::graphmap=0.6.3" : null)
-    container "quay.io/biocontainers/graphmap:0.6.3--he513fc3_0"
+    conda (params.enable_conda ? "bioconda::graphmap=0.6.3" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/graphmap:0.6.3--he513fc3_0"
+    } else {
+        container "quay.io/biocontainers/graphmap:0.6.3--he513fc3_0"
+    }

    input:
-    tuple val(meta), path(fastq)
-    path(fasta)
-    path(index)
+    tuple val(meta), path(reads)
+    path  fasta
+    path  index

    output:
-    tuple val(meta), path("*.sam"), emit: align_sam
+    tuple val(meta), path("*.sam"), emit: sam
+    path "*.version.txt"          , emit: version

    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
    """
    graphmap2 \\
        align \\
        -t $task.cpus \\
        -r $fasta \\
        -i $index \\
-        -d $fastq \\
-        -o ${meta.id}.sam \\
-        --extcigar
+        -d $reads \\
+        -o ${prefix}.sam \\
+        $options.args
+
+    echo \$(graphmap2 align 2>&1) | sed 's/^.*Version: v//; s/ .*\$//' > ${software}.version.txt
    """
 }
--- a/software/graphmap2/align/meta.yml
+++ b/software/graphmap2/align/meta.yml
@ -10,8 +10,8 @@ tools:
    - graphmap2:
        description: |
            A versatile pairwise aligner for genomic and spliced nucleotide sequences.
-        homepage: https://github.com/lh3/minimap2
-        documentation: https://github.com/lh3/minimap2#uguide
+        homepage: https://github.com/lbcb-sci/graphmap2
+        documentation: https://github.com/lbcb-sci/graphmap2#graphmap2---a-highly-sensitive-and-accurate-mapper-for-long-error-prone-reads
 input:
    - meta:
        type: map
@ -47,3 +47,4 @@ output:
        pattern: "*.{version.txt}"
 authors:
    - "@yuukiiwa"
+    - "@drpatelh"
--- a/software/graphmap2/index/main.nf
+++ b/software/graphmap2/index/main.nf
@ -2,30 +2,38 @@
 include { initOptions; saveFiles; getSoftwareName } from './functions'

 params.options = [:]
-def options    = initOptions(params.options)
+options        = initOptions(params.options)

 process GRAPHMAP2_INDEX {
+    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:'') }
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:[:], publish_by_meta:['']) }

-    conda     (params.enable_conda ? "bioconda::graphmap=0.6.3" : null)
-    container "quay.io/biocontainers/graphmap:0.6.3--he513fc3_0"
+    conda (params.enable_conda ? "bioconda::graphmap=0.6.3" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/graphmap:0.6.3--he513fc3_0"
+    } else {
+        container "quay.io/biocontainers/graphmap:0.6.3--he513fc3_0"
+    }

    input:
-    path(fasta)
+    path fasta

    output:
-    path("*.gmidx")       ,emit: index
-    path "*.version.txt"  ,emit: version
+    path "*.gmidx"      , emit: index
+    path "*.version.txt", emit: version

    script:
+    def software = getSoftwareName(task.process)
    """
    graphmap2 \\
        align \\
        -t $task.cpus \\
        -I \\
+        $options.args \\
        -r $fasta
-    echo \$(graphmap2 2>&1) > graphmap2.version.txt
+
+    echo \$(graphmap2 align 2>&1) | sed 's/^.*Version: v//; s/ .*\$//' > ${software}.version.txt
    """
 }
--- a/software/graphmap2/index/meta.yml
+++ b/software/graphmap2/index/meta.yml
@ -26,3 +26,4 @@ output:
        pattern: "*.{version.txt}"
 authors:
    - "@yuukiiwa"
+    - "@drpatelh"
--- a/software/minimap2/index/main.nf
+++ b/software/minimap2/index/main.nf
@ -2,31 +2,37 @@
 include { initOptions; saveFiles; getSoftwareName } from './functions'

 params.options = [:]
-def options    = initOptions(params.options)
+options        = initOptions(params.options)

 process MINIMAP2_INDEX {
    label 'process_medium'
    publishDir "${params.outdir}",
        mode: params.publish_dir_mode,
-        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['']) }
+        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:[:], publish_by_meta:['']) }

-    conda     (params.enable_conda ? "bioconda::minimap2=2.17" : null)
-    container "quay.io/biocontainers/minimap2:2.17--hed695b0_3"
+    conda (params.enable_conda ? "bioconda::minimap2=2.17" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/minimap2:2.17--hed695b0_3"
+    } else {
+        container "quay.io/biocontainers/minimap2:2.17--hed695b0_3"
+    }

    input:
-    path(fasta)
+    path fasta

    output:
-    path("*.mmi")        , emit: index
-    path "*.version.txt" ,emit: version
+    path "*.mmi"        , emit: index
+    path "*.version.txt", emit: version

    script:
+    def software = getSoftwareName(task.process)
    """
    minimap2 \\
        -t $task.cpus \\
-        -d ${fasta}.mmi \\
+        -d ${fasta.baseName}.mmi \\
+        $options.args \\
        $fasta
-    ps
-    minimap2 --version &> minimap2.version.txt
+
+    echo \$(minimap2 --version 2>&1) > ${software}.version.txt
    """
 }
--- a/software/minimap2/index/meta.yml
+++ b/software/minimap2/index/meta.yml
@ -26,3 +26,4 @@ output:
        pattern: "*.{version.txt}"
 authors:
    - "@yuukiiwa"
+    - "@drpatelh"
--- a/software/qcat/main.nf
+++ b/software/qcat/main.nf
@ -2,7 +2,7 @@
 include { initOptions; saveFiles; getSoftwareName } from './functions'

 params.options = [:]
-def options    = initOptions(params.options)
+options        = initOptions(params.options)

 process QCAT {
    tag "$meta.id"
@ -11,34 +11,42 @@ process QCAT {
        mode: params.publish_dir_mode,
        saveAs: { filename -> saveFiles(filename:filename, options:params.options, publish_dir:getSoftwareName(task.process), meta:meta, publish_by_meta:['id']) }

-    conda     (params.enable_conda ? "bioconda::qcat=1.1.0" : null)
-    container "quay.io/biocontainers/qcat:1.1.0--py_0"
+    conda (params.enable_conda ? "bioconda::qcat=1.1.0" : null)
+    if (workflow.containerEngine == 'singularity' && !params.singularity_pull_docker_container) {
+        container "https://depot.galaxyproject.org/singularity/qcat:1.1.0--py_0"
+    } else {
+        container "quay.io/biocontainers/qcat:1.1.0--py_0"
+    }

    input:
-    tuple val(meta), path(input_path)
-    val(barcode_kit)
+    tuple val(meta), path(reads)
+    val   barcode_kit

    output:
-    tuple val(meta), path("fastq/*.fastq.gz") , emit: fastq
-    path "*.version.txt"                      , emit: version
+    tuple val(meta), path("fastq/*.fastq.gz"), emit: reads
+    path "*.version.txt"                     , emit: version

    script:
+    def software = getSoftwareName(task.process)
+    def prefix   = options.suffix ? "${meta.id}${options.suffix}" : "${meta.id}"
    """
    ## Unzip fastq file
-    ## qcat doesnt support zipped files yet
-    FILE=$input_path
+    ## qcat doesn't support zipped files yet
+    FILE=$reads
    if [[ \$FILE == *.gz ]]
    then
-    zcat $input_path > unzipped.fastq
-    FILE=unzipped.fastq
+        zcat $reads > unzipped.fastq
+        FILE=unzipped.fastq
    fi
-    qcat  \\
-    -f \$FILE \\
-    -b ./fastq \\
-    --kit $barcode_kit

-    ## Zip fastq files (cannot find pigz command)
+    qcat \\
+        -f \$FILE \\
+        -b ./fastq \\
+        --kit $barcode_kit
+
+    ## Zip fastq files
    gzip fastq/*
-    qcat --version &> qcat.version.txt
+
+    echo \$(qcat --version 2>&1) | sed 's/^.*qcat //; s/ .*\$//' > ${software}.version.txt
    """
 }
--- a/software/qcat/meta.yml
+++ b/software/qcat/meta.yml
@ -15,10 +15,10 @@ input:
        description: |
            Groovy Map containing sample information
            e.g. [ id:'test', single_end:false ]
-    - input_path:
+    - reads:
        type: file
        description: |
-            Non-demultiplexed Nanopore sequencing sample.
+            Non-demultiplexed fastq files

 output:
    - meta:
@ -26,7 +26,7 @@ output:
        description: |
            Groovy Map containing sample information
            e.g. [ id:'test', single_end:false ]
-    - fastq:
+    - reads:
        type: file
        description: Demultiplexed fastq samples
        pattern: "*.fastq.gz"
@ -36,3 +36,4 @@ output:
        pattern: "*.{version.txt}"
 authors:
    - "@yuukiiwa"
+    - "@drpatelh"
--- a/tests/config/test_data.config
+++ b/tests/config/test_data.config
@ -146,9 +146,6 @@ params {
                test2_genome_vcf_gz_tbi                        = "${test_data_dir}/genomics/homo_sapiens/illumina/gvcf/test2.genome.vcf.gz.tbi"
                test2_genome_vcf_idx                           = "${test_data_dir}/genomics/homo_sapiens/illumina/gvcf/test2.genome.vcf.idx"
            }
-            'nanopore' {
-                non_demultiplexed_fastq                        = "https://github.com/nf-core/test-datasets/raw/nanoseq/fastq/nondemultiplexed/sample_nobc_dx.fastq.gz"
-            }
        }
    }
 }
--- a/tests/software/bedtools/bamtobed/main.nf
+++ b/tests/software/bedtools/bamtobed/main.nf
@ -6,9 +6,7 @@ include { BEDTOOLS_BAMTOBED } from '../../../../software/bedtools/bamtobed/main.

 workflow test_bedtools_bamtobed {
    input = [ [ id:'test'], //meta map
-              file(params.test_data['sarscov2']['genome']['genome_sizes'], checkIfExists: true),
-              file(params.test_data['sarscov2']['nanopore']['test_sorted_bam'], checkIfExists: true),
-              file(params.test_data['sarscov2']['nanopore']['test_sorted_bam_bai'], checkIfExists: true)
+              file(params.test_data['sarscov2']['illumina']['test_single_end_bam'], checkIfExists: true)
            ]

    BEDTOOLS_BAMTOBED ( input )
--- a/tests/software/bedtools/bamtobed/test.yml
+++ b/tests/software/bedtools/bamtobed/test.yml
@ -4,5 +4,5 @@
    - bedtools
    - bedtools/bamtobed
  files:
-    - path: ./output/bedtools/test.bed12
-      md5sum: d41d8cd98f00b204e9800998ecf8427e
+    - path: ./output/bedtools/test.bed
+      md5sum: 3c6b88e414debd889c10eb972180b687
--- a/tests/software/minimap2/index/test.yml
+++ b/tests/software/minimap2/index/test.yml
@ -4,5 +4,5 @@
    - minimap2
    - minimap2/index
  files:
-    - path: ./output/minimap2/genome.fasta.mmi
+    - path: ./output/minimap2/genome.mmi
      md5sum: 72e450f12dc691e763c697463bdb1571
--- a/tests/software/qcat/main.nf
+++ b/tests/software/qcat/main.nf
@ -2,13 +2,13 @@

 nextflow.enable.dsl = 2

-include { QCAT  } from '../../../software/qcat/main.nf' addParams( options: [:] )
+include { QCAT } from '../../../software/qcat/main.nf' addParams( options: [:] )

 workflow test_qcat {
-    def input = []
    input = [ [ id:'test' ], // meta map
-            [ file(params.test_data['homo_sapiens']['nanopore']['non_demultiplexed_fastq'], checkIfExists: true) ]
+              [ file("https://github.com/nf-core/test-datasets/raw/nanoseq/fastq/nondemultiplexed/sample_nobc_dx.fastq.gz", checkIfExists: true) ]
            ]
    barcode_kit = 'NBD103/NBD104'
+
    QCAT ( input, barcode_kit )
 }
--- a/tests/software/qcat/test.yml
+++ b/tests/software/qcat/test.yml
@ -5,4 +5,4 @@
  files:
    - path: ./output/qcat/fastq/barcode06.fastq.gz
    - path: ./output/qcat/fastq/barcode12.fastq.gz
-    - path: ./output/qcat/fastq/none.fastq.gz
+    - path: ./output/qcat/fastq/none.fastq.gz