Merge branch 'new-module-rtg/vcfeval' of github.com:nvnieuwk/modules into new-module-rtg/vcfeval

modules/meryl/histogram/main.nf

@@ -0,0 +1,34 @@
+process MERYL_HISTOGRAM {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::meryl=1.3" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/meryl:1.3--h87f3376_1':
+        'quay.io/biocontainers/meryl:1.3--h87f3376_1' }"
+
+    input:
+    tuple val(meta), path(meryl_db)
+
+    output:
+    tuple val(meta), path("*.hist"), emit: hist
+    path "versions.yml"            , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    meryl histogram \\
+        threads=$task.cpus \\
+        $args \\
+        $meryl_db > ${prefix}.hist
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        meryl: \$( meryl --version |& sed 's/meryl //' )
+    END_VERSIONS
+    """
+}

modules/meryl/histogram/meta.yml

@@ -0,0 +1,41 @@
+name: "meryl_histogram"
+description: A genomic k-mer counter (and sequence utility) with nice features.
+keywords:
+  - k-mer
+  - histogram
+tools:
+  - "meryl":
+      description: "A genomic k-mer counter (and sequence utility) with nice features. "
+      homepage: "https://github.com/marbl/meryl"
+      documentation: "https://meryl.readthedocs.io/en/latest/quick-start.html"
+      tool_dev_url: "https://github.com/marbl/meryl"
+      doi: ""
+      licence: "['GPL']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - meryl_dbs:
+      type: directory
+      description: Meryl k-mer database
+
+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"
+  - hist:
+      type: file
+      description: Histogram of k-mers
+      pattern: "*.hist"
+
+authors:
+  - "@mahesh-panchal"

modules/meryl/unionsum/main.nf

@@ -0,0 +1,35 @@
+process MERYL_UNIONSUM {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::meryl=1.3" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/meryl:1.3--h87f3376_1':
+        'quay.io/biocontainers/meryl:1.3--h87f3376_1' }"
+
+    input:
+    tuple val(meta), path(meryl_dbs)
+
+    output:
+    tuple val(meta), path("*.unionsum.meryldb"), emit: meryl_db
+    path "versions.yml"                        , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    meryl union-sum \\
+        threads=$task.cpus \\
+        $args \\
+        output ${prefix}.unionsum.meryldb \\
+        $meryl_dbs
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        meryl: \$( meryl --version |& sed 's/meryl //' )
+    END_VERSIONS
+    """
+}

modules/meryl/unionsum/meta.yml

@@ -0,0 +1,41 @@
+name: "meryl_unionsum"
+description: A genomic k-mer counter (and sequence utility) with nice features.
+keywords:
+  - k-mer
+  - unionsum
+tools:
+  - "meryl":
+      description: "A genomic k-mer counter (and sequence utility) with nice features. "
+      homepage: "https://github.com/marbl/meryl"
+      documentation: "https://meryl.readthedocs.io/en/latest/quick-start.html"
+      tool_dev_url: "https://github.com/marbl/meryl"
+      doi: ""
+      licence: "['GPL']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - meryl_dbs:
+      type: directory
+      description: Meryl k-mer databases
+
+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"
+  - meryl_db:
+      type: directory
+      description: A Meryl k-mer database that is the union sum of the input databases
+      pattern: "*.unionsum.meryldb"
+
+authors:
+  - "@mahesh-panchal"

modules/samtools/convert/main.nf

@@ -1,5 +1,4 @@
-//There is a -L option to only output alignments in interval, might be an option for exons/panel data?
-process SAMTOOLS_BAMTOCRAM {
+process SAMTOOLS_CONVERT {
     tag "$meta.id"
     label 'process_medium'
 
@@ -14,8 +13,8 @@ process SAMTOOLS_BAMTOCRAM {
     path  fai
 
     output:
-    tuple val(meta), path("*.cram"), path("*.crai"), emit: cram_crai
-    path  "versions.yml"                           , emit: versions
+    tuple val(meta), path("*.{cram,bam}"), path("*.{crai,bai}") , emit: alignment_index
+    path  "versions.yml"                                        , emit: versions
 
     when:
     task.ext.when == null || task.ext.when
@@ -23,9 +22,17 @@ process SAMTOOLS_BAMTOCRAM {
     script:
     def args = task.ext.args  ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
+    def output_extension = input.getExtension() == "bam" ? "cram" : "bam"
+
     """
-    samtools view --threads ${task.cpus} --reference ${fasta} -C $args $input > ${prefix}.cram
-    samtools index -@${task.cpus} ${prefix}.cram
+    samtools view \\
+        --threads ${task.cpus} \\
+        --reference ${fasta} \\
+        $args \\
+        $input \\
+        -o ${prefix}.${output_extension}
+
+    samtools index -@${task.cpus} ${prefix}.${output_extension}
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":

modules/samtools/convert/meta.yml

@@ -1,5 +1,5 @@
-name: samtools_bamtocram
-description: filter/convert and then index CRAM file
+name: samtools_convert
+description: convert and then index CRAM -> BAM or BAM -> CRAM file
 keywords:
   - view
   - index
@@ -23,12 +23,12 @@ input:
         e.g. [ id:'test', single_end:false ]
   - input:
       type: file
-      description: BAM/SAM file
-      pattern: "*.{bam,sam}"
+      description: BAM/CRAM file
+      pattern: "*.{bam,cram}"
   - index:
       type: file
-      description: BAM/SAM index file
-      pattern: "*.{bai,sai}"
+      description: BAM/CRAM index file
+      pattern: "*.{bai,crai}"
   - fasta:
       type: file
       description: Reference file to create the CRAM file
@@ -39,10 +39,10 @@ output:
       description: |
         Groovy Map containing sample information
         e.g. [ id:'test', single_end:false ]
-  - cram_crai:
+  - alignment_index:
       type: file
-      description: filtered/converted CRAM file + index
-      pattern: "*{.cram,.crai}"
+      description: filtered/converted BAM/CRAM file + index
+      pattern: "*{.bam/cram,.bai/crai}"
   - version:
       type: file
       description: File containing software version

tests/config/pytest_modules.yml

@@ -1218,6 +1218,14 @@ meryl/count:
   - modules/meryl/count/**
   - tests/modules/meryl/count/**
 
+meryl/histogram:
+  - modules/meryl/histogram/**
+  - tests/modules/meryl/histogram/**
+
+meryl/unionsum:
+  - modules/meryl/unionsum/**
+  - tests/modules/meryl/unionsum/**
+
 metabat2/jgisummarizebamcontigdepths:
   - modules/metabat2/jgisummarizebamcontigdepths/**
   - tests/modules/metabat2/jgisummarizebamcontigdepths/**
@@ -1635,9 +1643,9 @@ samtools/bam2fq:
   - modules/samtools/bam2fq/**
   - tests/modules/samtools/bam2fq/**
 
-samtools/bamtocram:
-  - modules/samtools/bamtocram/**
-  - tests/modules/samtools/bamtocram/**
+samtools/convert:
+  - modules/samtools/convert/**
+  - tests/modules/samtools/convert/**
 
 samtools/collatefastq:
   - modules/samtools/collatefastq/**
@@ -1771,14 +1779,14 @@ slimfastq:
   - modules/slimfastq/**
   - tests/modules/slimfastq/**
 
-snapaligner/index:
-  - modules/snapaligner/index/**
-  - tests/modules/snapaligner/index/**
-
 snapaligner/align:
   - modules/snapaligner/align/**
   - tests/modules/snapaligner/align/**
 
+snapaligner/index:
+  - modules/snapaligner/index/**
+  - tests/modules/snapaligner/index/**
+
 snpdists:
   - modules/snpdists/**
   - tests/modules/snpdists/**

tests/modules/meryl/histogram/main.nf

@@ -0,0 +1,17 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { MERYL_COUNT     } from '../../../../modules/meryl/count/main.nf'
+include { MERYL_HISTOGRAM } from '../../../../modules/meryl/histogram/main.nf'
+
+workflow test_meryl_histogram {
+
+    input = [
+        [ id:'test' ], // meta map
+        file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+    ]
+
+    MERYL_COUNT ( input )
+    MERYL_HISTOGRAM ( MERYL_COUNT.out.meryl_db )
+}

tests/modules/meryl/histogram/nextflow.config

@@ -1,5 +1,6 @@
 process {
 
     publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    ext.args = 'k=21'
 
 }

tests/modules/meryl/histogram/test.yml

@@ -0,0 +1,10 @@
+- name: meryl histogram test_meryl_histogram
+  command: nextflow run tests/modules/meryl/histogram -entry test_meryl_histogram -c tests/config/nextflow.config
+  tags:
+    - meryl/histogram
+    - meryl
+  files:
+    - path: output/meryl/test.hist
+      md5sum: 4bfdc8b287ee0cfd9922bbfa8cd64650
+    - path: output/meryl/versions.yml
+      md5sum: 050038f1b1df79977a393cce1b4b2ddb

tests/modules/meryl/unionsum/main.nf

@@ -0,0 +1,31 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { MERYL_COUNT    } from '../../../../modules/meryl/count/main.nf'
+include { MERYL_UNIONSUM } from '../../../../modules/meryl/unionsum/main.nf'
+
+workflow test_meryl_unionsum_single_end {
+
+    input = [
+        [ id:'test', single_end: true ], // meta map
+        file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+    ]
+
+    MERYL_COUNT ( input )
+    MERYL_UNIONSUM ( MERYL_COUNT.out.meryl_db )
+}
+
+workflow test_meryl_unionsum_paired_end {
+
+    input = [
+        [ id:'test', single_end: false ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['sarscov2']['illumina']['test_2_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+
+    MERYL_COUNT ( input )
+    MERYL_UNIONSUM ( MERYL_COUNT.out.meryl_db )
+}

tests/modules/meryl/unionsum/nextflow.config

@@ -0,0 +1,6 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    ext.args = 'k=21'
+
+}

tests/modules/meryl/unionsum/test.yml

@@ -0,0 +1,17 @@
+- name: meryl unionsum test_meryl_unionsum_single_end
+  command: nextflow run tests/modules/meryl/unionsum -entry test_meryl_unionsum_single_end -c tests/config/nextflow.config
+  tags:
+    - meryl
+    - meryl/unionsum
+  files:
+    - path: output/meryl/versions.yml
+      md5sum: 7de859c6d3a29d72f6c9c976609d0913
+
+- name: meryl unionsum test_meryl_unionsum_paired_end
+  command: nextflow run tests/modules/meryl/unionsum -entry test_meryl_unionsum_paired_end -c tests/config/nextflow.config
+  tags:
+    - meryl
+    - meryl/unionsum
+  files:
+    - path: output/meryl/versions.yml
+      md5sum: a16decdec014ccb9bdab69a4a1d30818

tests/modules/samtools/bamtocram/main.nf

@@ -1,17 +0,0 @@
-#!/usr/bin/env nextflow
-
-nextflow.enable.dsl = 2
-
-include { SAMTOOLS_BAMTOCRAM } from '../../../../modules/samtools/bamtocram/main.nf'
-
-workflow test_samtools_bamtocram {
-
-    input = [ [ id:'test', single_end:false ], // meta map
-                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
-                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
-
-    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
-    fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
-
-    SAMTOOLS_BAMTOCRAM ( input, fasta, fai )
-}

tests/modules/samtools/bamtocram/test.yml

@@ -1,9 +0,0 @@
-- name: samtools bamtocram test_samtools_bamtocram
-  command: nextflow run ./tests/modules/samtools/bamtocram -entry test_samtools_bamtocram -c ./tests/config/nextflow.config -c ./tests/modules/samtools/bamtocram/nextflow.config
-  tags:
-    - samtools/bamtocram
-    - samtools
-  files:
-    - path: output/samtools/test.cram
-    - path: output/samtools/test.cram.crai
-    - path: output/samtools/versions.yml

tests/modules/samtools/convert/main.nf

@@ -0,0 +1,31 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { SAMTOOLS_CONVERT as SAMTOOLS_BAMTOCRAM } from '../../../../modules/samtools/convert/main.nf'
+include { SAMTOOLS_CONVERT as SAMTOOLS_CRAMTOBAM } from '../../../../modules/samtools/convert/main.nf'
+
+workflow test_samtools_convert_bamtocram {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam'], checkIfExists: true),
+                file(params.test_data['sarscov2']['illumina']['test_paired_end_sorted_bam_bai'], checkIfExists: true)]
+
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    fai   = file(params.test_data['sarscov2']['genome']['genome_fasta_fai'], checkIfExists: true)
+
+    SAMTOOLS_BAMTOCRAM ( input, fasta, fai )
+}
+
+workflow test_samtools_convert_cramtobam {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram'], checkIfExists: true),
+                file(params.test_data['homo_sapiens']['illumina']['test_paired_end_recalibrated_sorted_cram_crai'], checkIfExists: true)
+            ]
+
+    fasta = file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    fai   = file(params.test_data['homo_sapiens']['genome']['genome_fasta_fai'], checkIfExists: true)
+
+    SAMTOOLS_CRAMTOBAM ( input, fasta, fai )
+}

tests/modules/samtools/convert/nextflow.config

@@ -0,0 +1,12 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName:SAMTOOLS_BAMTOCRAM{
+        ext.args = "-C"
+    }
+
+    withName:SAMTOOLS_CRAMTOBAM{
+        ext.args = "-b"
+    }
+}

tests/modules/samtools/convert/test.yml

@@ -0,0 +1,21 @@
+- name: samtools convert test_samtools_convert_bamtocram
+  command: nextflow run tests/modules/samtools/convert -entry test_samtools_convert_bamtocram -c tests/config/nextflow.config
+  tags:
+    - samtools
+    - samtools/convert
+  files:
+    - path: output/samtools/test.cram
+    - path: output/samtools/test.cram.crai
+    - path: output/samtools/versions.yml
+
+- name: samtools convert test_samtools_convert_cramtobam
+  command: nextflow run tests/modules/samtools/convert -entry test_samtools_convert_cramtobam -c tests/config/nextflow.config
+  tags:
+    - samtools
+    - samtools/convert
+  files:
+    - path: output/samtools/test.bam
+      md5sum: c262b6dc15f9b480bdb47d6d018b4b56
+    - path: output/samtools/test.bam.bai
+      md5sum: 6e8f5034f728401bfa841c8e70c62463
+    - path: output/samtools/versions.yml