Merge pull request #1641 from mahesh-panchal/genomescope2

Genomescope2 module

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
+    """
+}

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"

tests/config/pytest_modules.yml

@@ -839,6 +839,10 @@ genmap/mappability:
   - modules/genmap/mappability/**
   - tests/modules/genmap/mappability/**
 
+genomescope2:
+  - modules/genomescope2/**
+  - tests/modules/genomescope2/**
+
 genrich:
   - modules/genrich/**
   - tests/modules/genrich/**
@@ -1647,14 +1651,14 @@ samtools/bam2fq:
   - modules/samtools/bam2fq/**
   - tests/modules/samtools/bam2fq/**
 
-samtools/convert:
-  - modules/samtools/convert/**
-  - tests/modules/samtools/convert/**
-
 samtools/collatefastq:
   - modules/samtools/collatefastq/**
   - tests/modules/samtools/collatefastq/**
 
+samtools/convert:
+  - modules/samtools/convert/**
+  - tests/modules/samtools/convert/**
+
 samtools/depth:
   - modules/samtools/depth/**
   - tests/modules/samtools/depth/**

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 )
+}

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'
+    }
+
+}

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