Merge branch 'new-module-gatk4/splitintervals' of github.com:nvnieuwk/modules into new-module-gatk4/splitintervals

2024-11-14 05:43:08 +00:00 · 2022-05-13 13:48:35 +02:00 · 2022-05-13 13:48:35 +02:00 · 4f6264f7c3
commit 4f6264f7c3
parent 9d53577282 1d4862ba60
10 changed files with 224 additions and 12 deletions
--- a/modules/bowtie2/align/main.nf
+++ b/modules/bowtie2/align/main.nf
@ -69,4 +69,3 @@ process BOWTIE2_ALIGN {
    END_VERSIONS
    """
 }
-
--- a/modules/genomescope2/main.nf
+++ b/modules/genomescope2/main.nf
@ -0,0 +1,40 @@
+process GENOMESCOPE2 {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::genomescope2=2.0" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/genomescope2:2.0--py310r41hdfd78af_5':
+        'quay.io/biocontainers/genomescope2:2.0--py310r41hdfd78af_5' }"
+
+    input:
+    tuple val(meta), path(histogram)
+
+    output:
+    tuple val(meta), path("*_linear_plot.png")            , emit: linear_plot_png
+    tuple val(meta), path("*_transformed_linear_plot.png"), emit: transformed_linear_plot_png
+    tuple val(meta), path("*_log_plot.png")               , emit: log_plot_png
+    tuple val(meta), path("*_transformed_log_plot.png")   , emit: transformed_log_plot_png
+    tuple val(meta), path("*_model.txt")                  , emit: model
+    tuple val(meta), path("*_summary.txt")                , emit: summary
+    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}"
+    """
+    genomescope2 \\
+        --input $histogram \\
+        $args \\
+        --output . \\
+        --name_prefix $prefix
+
+    cat <<-END_VERSIONS > versions.yml
+    '${task.process}':
+        genomescope2: \$( genomescope2 -v | sed 's/GenomeScope //' )
+    END_VERSIONS
+    """
+}
--- a/modules/genomescope2/meta.yml
+++ b/modules/genomescope2/meta.yml
@ -0,0 +1,67 @@
+name: "genomescope2"
+description: Estimate genome heterozygosity, repeat content, and size from sequencing reads using a kmer-based statistical approach
+keywords:
+  - "genome size"
+  - "genome heterozygosity"
+  - "repeat content"
+tools:
+  - "genomescope2":
+      description: "Reference-free profiling of polyploid genomes"
+      homepage: "http://qb.cshl.edu/genomescope/genomescope2.0/"
+      documentation: "https://github.com/tbenavi1/genomescope2.0/blob/master/README.md"
+      tool_dev_url: "https://github.com/tbenavi1/genomescope2.0"
+      doi: "https://doi.org/10.1038/s41467-020-14998-3"
+      licence: "['Apache License, Version 2.0 (Apache-2.0)']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - histogram:
+      type: file
+      description: A K-mer histogram file
+      pattern: "*.hist"
+
+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"
+  - linear_plot_png:
+      type: file
+      description: A genomescope2 linear plot in PNG format
+      pattern: "*_linear_plot.png"
+  - linear_plot_png:
+      type: file
+      description: A genomescope2 linear plot in PNG format
+      pattern: "*_linear_plot.png"
+  - transformed_linear_plot_png:
+      type: file
+      description: A genomescope2 transformed linear plot in PNG format
+      pattern: "*_transformed_linear_plot.png"
+  - log_plot_png:
+      type: file
+      description: A genomescope2 log plot in PNG format
+      pattern: "*_log_plot.png"
+  - transformed_log_plot_png:
+      type: file
+      description: A genomescope2 transformed log plot in PNG format
+      pattern: "*_transformed_log_plot.png"
+  - model:
+      type: file
+      description: Genomescope2 model fit summary
+      pattern: "*_model.txt"
+  - summary:
+      type: file
+      description: Genomescope2 histogram summary
+      pattern: "*_summary.txt"
+
+authors:
+  - "@mahesh-panchal"
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@ -843,6 +843,10 @@ genmap/mappability:
  - modules/genmap/mappability/**
  - tests/modules/genmap/mappability/**

+genomescope2:
+  - modules/genomescope2/**
+  - tests/modules/genomescope2/**
+
 genrich:
  - modules/genrich/**
  - tests/modules/genrich/**
--- a/tests/modules/bowtie2/align/main.nf
+++ b/tests/modules/bowtie2/align/main.nf
@ -33,3 +33,32 @@ workflow test_bowtie2_align_paired_end {
    BOWTIE2_BUILD ( fasta )
    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
 }
+
+workflow test_bowtie2_align_single_end_large_index {
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        [
+            file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    save_unaligned = false
+
+    BOWTIE2_BUILD ( fasta )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
+}
+
+workflow test_bowtie2_align_paired_end_large_index {
+    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)
+        ]
+    ]
+    fasta = file(params.test_data['sarscov2']['genome']['genome_fasta'], checkIfExists: true)
+    save_unaligned = false
+
+    BOWTIE2_BUILD ( fasta )
+    BOWTIE2_ALIGN ( input, BOWTIE2_BUILD.out.index, save_unaligned )
+}
--- a/tests/modules/bowtie2/align/nextflow.config
+++ b/tests/modules/bowtie2/align/nextflow.config
@ -5,6 +5,7 @@ params {
 process {

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

 if (params.force_large_index) {
--- a/tests/modules/bowtie2/align/test.yml
+++ b/tests/modules/bowtie2/align/test.yml
@ -1,21 +1,39 @@
 - name: bowtie2 align test_bowtie2_align_single_end
  command: nextflow run ./tests/modules/bowtie2/align -entry test_bowtie2_align_single_end -c ./tests/config/nextflow.config -c ./tests/modules/bowtie2/align/nextflow.config
  tags:
-    - bowtie2/align
    - bowtie2
+    - bowtie2/align
  files:
-    - path: output/bowtie2/test.bam
-    - path: output/bowtie2/test.bowtie2.log
-      md5sum: 7b8a9e61b7646da1089b041333c41a87
-    - path: output/bowtie2/versions.yml
+    - path: ./output/bowtie2/test.bam
+    - path: ./output/bowtie2/test.bowtie2.log
+    - path: ./output/bowtie2/versions.yml

 - name: bowtie2 align test_bowtie2_align_paired_end
-  command: nextflow run ./tests/modules/bowtie2/align -entry test_bowtie2_align_paired_end -c ./tests/config/nextflow.config -c ./tests/modules/bowtie2/align/nextflow.config
+  command: nextflow run tests/modules/bowtie2/align -entry test_bowtie2_align_paired_end -c tests/config/nextflow.config -c tests/modules/bowtie2/align/nextflow.config
  tags:
-    - bowtie2/align
    - bowtie2
+    - bowtie2/align
  files:
-    - path: output/bowtie2/test.bam
-    - path: output/bowtie2/test.bowtie2.log
-      md5sum: bd89ce1b28c93bf822bae391ffcedd19
-    - path: output/bowtie2/versions.yml
+    - path: ./output/bowtie2/test.bam
+    - path: ./output/bowtie2/test.bowtie2.log
+    - path: ./output/bowtie2/versions.yml
+
+- name: bowtie2 align test_bowtie2_align_single_end_large_index
+  command: nextflow run tests/modules/bowtie2/align -entry test_bowtie2_align_single_end_large_index -c tests/config/nextflow.config -c tests/modules/bowtie2/align/nextflow.config --force_large_index
+  tags:
+    - bowtie2
+    - bowtie2/align
+  files:
+    - path: ./output/bowtie2/test.bam
+    - path: ./output/bowtie2/test.bowtie2.log
+    - path: ./output/bowtie2/versions.yml
+
+- name: bowtie2 align test_bowtie2_align_paired_end_large_index
+  command: nextflow run tests/modules/bowtie2/align -entry test_bowtie2_align_paired_end_large_index -c tests/config/nextflow.config -c tests/modules/bowtie2/align/nextflow.config --force_large_index
+  tags:
+    - bowtie2
+    - bowtie2/align
+  files:
+    - path: ./output/bowtie2/test.bam
+    - path: ./output/bowtie2/test.bowtie2.log
+    - path: ./output/bowtie2/versions.yml
--- a/tests/modules/genomescope2/main.nf
+++ b/tests/modules/genomescope2/main.nf
@ -0,0 +1,19 @@
+#!/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'
+include { GENOMESCOPE2    } from '../../../modules/genomescope2/main.nf'
+
+workflow test_genomescope2 {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['bacteroides_fragilis']['illumina']['test1_1_fastq_gz'], checkIfExists: true)
+    ]
+
+    MERYL_COUNT ( input )
+    MERYL_HISTOGRAM ( MERYL_COUNT.out.meryl_db )
+    GENOMESCOPE2 ( MERYL_HISTOGRAM.out.hist )
+}
--- a/tests/modules/genomescope2/nextflow.config
+++ b/tests/modules/genomescope2/nextflow.config
@ -0,0 +1,13 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName: 'MERYL.*' {
+        ext.args = 'k=21'
+    }
+
+    withName: 'GENOMESCOPE2' {
+        ext.args = '-k 21 -p 1'
+    }
+
+}
--- a/tests/modules/genomescope2/test.yml
+++ b/tests/modules/genomescope2/test.yml
@ -0,0 +1,22 @@
+- name: genomescope2 test_genomescope2
+  command: nextflow run tests/modules/genomescope2 -entry test_genomescope2 -c tests/config/nextflow.config
+  tags:
+    - genomescope2
+  files:
+    - path: output/genomescope2/test_linear_plot.png
+      md5sum: 94c165c5028156299a1d4d05766cac51
+    - path: output/genomescope2/test_log_plot.png
+      md5sum: 9d25ca463d92a0c73a893da7fd3979ba
+    - path: output/genomescope2/test_model.txt
+      md5sum: 3caf62f715f64a2f2b8fdff5d079cb84
+    - path: output/genomescope2/test_summary.txt
+      md5sum: 7452860e2cea99b85f3ff60daeac77f5
+    - path: output/genomescope2/test_transformed_linear_plot.png
+      md5sum: 99a64c1c18d8670f64cb863d4334abbb
+    - path: output/genomescope2/test_transformed_log_plot.png
+      md5sum: b4e029c9fb9987ca33b17392a691c1b4
+    - path: output/genomescope2/versions.yml
+      md5sum: 18afeb26f62a47f680b2bb3e27da9cbc
+    - path: output/meryl/test.hist
+      md5sum: f75362ab9cd70d96621b3690e952085f
+    - path: output/meryl/versions.yml