Add Merqury (#1647)

* Add merqury files

* Remove md5sum for completeness stats file

* update container and test output

* Update modules/merqury/meta.yml

Co-authored-by: Daniel Straub <42973691+d4straub@users.noreply.github.com>

* Explicitly set thread usage for meryl

* Remove empty file md5sum

Co-authored-by: Daniel Straub <42973691+d4straub@users.noreply.github.com>

modules/merqury/main.nf

@@ -0,0 +1,59 @@
+process MERQURY {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::merqury=1.3" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/merqury:1.3--hdfd78af_1':
+        'quay.io/biocontainers/merqury:1.3--hdfd78af_1' }"
+
+    input:
+    tuple val(meta), path(meryl_db), path(assembly)
+
+    output:
+    tuple val(meta), path("*_only.bed")          , emit: assembly_only_kmers_bed
+    tuple val(meta), path("*_only.wig")          , emit: assembly_only_kmers_wig
+    tuple val(meta), path("*.completeness.stats"), emit: stats
+    tuple val(meta), path("*.dist_only.hist")    , emit: dist_hist
+    tuple val(meta), path("*.spectra-cn.fl.png") , emit: spectra_cn_fl_png
+    tuple val(meta), path("*.spectra-cn.hist")   , emit: spectra_cn_hist
+    tuple val(meta), path("*.spectra-cn.ln.png") , emit: spectra_cn_ln_png
+    tuple val(meta), path("*.spectra-cn.st.png") , emit: spectra_cn_st_png
+    tuple val(meta), path("*.spectra-asm.fl.png"), emit: spectra_asm_fl_png
+    tuple val(meta), path("*.spectra-asm.hist")  , emit: spectra_asm_hist
+    tuple val(meta), path("*.spectra-asm.ln.png"), emit: spectra_asm_ln_png
+    tuple val(meta), path("*.spectra-asm.st.png"), emit: spectra_asm_st_png
+    tuple val(meta), path("${prefix}.qv")        , emit: assembly_qv
+    tuple val(meta), path("${prefix}.*.qv")      , emit: scaffold_qv
+    tuple val(meta), path("*.hist.ploidy")       , emit: read_ploidy
+    path "versions.yml"                          , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    // def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def VERSION = 1.3
+    """
+    # Nextflow changes the container --entrypoint to /bin/bash (container default entrypoint: /usr/local/env-execute)
+    # Check for container variable initialisation script and source it.
+    if [ -f "/usr/local/env-activate.sh" ]; then
+        set +u  # Otherwise, errors out because of various unbound variables
+        . "/usr/local/env-activate.sh"
+        set -u
+    fi
+    # limit meryl to use the assigned number of cores.
+    export OMP_NUM_THREADS=$task.cpus
+
+    merqury.sh \\
+        $meryl_db \\
+        $assembly \\
+        $prefix
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        merqury: $VERSION
+    END_VERSIONS
+    """
+}

modules/merqury/meta.yml

@@ -0,0 +1,101 @@
+name: "merqury"
+description: k-mer based assembly evaluation.
+keywords:
+  - "k-mer"
+  - "assembly"
+  - "evaluation"
+tools:
+  - "merqury":
+      description: "Evaluate genome assemblies with k-mers and more."
+      homepage: "None"
+      documentation: "None"
+      tool_dev_url: "https://github.com/marbl/merqury"
+      doi: "10.1186/s13059-020-02134-9"
+      licence: "['PUBLIC DOMAIN']"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - meryl_db:
+      type: file
+      description: "Meryl read database"
+  - assembly:
+      type: file
+      description: FASTA assembly file
+
+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"
+  - assembly_only_kmers_bed:
+      type: file
+      description: "The positions of the k-mers found only in an assembly for further investigation in .bed"
+      pattern: "*_only.bed"
+  - assembly_only_kmers_wig:
+      type: file
+      description: "The positions of the k-mers found only in an assembly for further investigation in .wig"
+      pattern: "*_only.wig"
+  - stats:
+      type: file
+      description: Assembly statistics file
+      pattern: "*.completeness.stats"
+  - dist_hist:
+      type: file
+      description: Histogram
+      pattern: "*.dist_only.hist"
+  - spectra_cn_fl_png:
+      type: file
+      description: "Unstacked copy number spectra filled plot in PNG format"
+      pattern: "*.spectra-cn.fl.png"
+  - spectra_cn_ln_png:
+      type: file
+      description: "Unstacked copy number spectra line plot in PNG format"
+      pattern: "*.spectra-cn.ln.png"
+  - spectra_cn_st_png:
+      type: file
+      description: "Stacked copy number spectra line plot in PNG format"
+      pattern: "*.spectra-cn.st.png"
+  - spectra_cn_hist:
+      type: file
+      description: "Copy number spectra histogram"
+      pattern: "*.spectra-cn.hist"
+  - spectra_asm_fl_png:
+      type: file
+      description: "Unstacked assembly spectra filled plot in PNG format"
+      pattern: "*.spectra-asm.fl.png"
+  - spectra_asm_ln_png:
+      type: file
+      description: "Unstacked assembly spectra line plot in PNG format"
+      pattern: "*.spectra-asm.ln.png"
+  - spectra_asm_st_png:
+      type: file
+      description: "Stacked assembly spectra line plot in PNG format"
+      pattern: "*.spectra-asm.st.png"
+  - spectra_asm_hist:
+      type: file
+      description: "Assembly spectra histogram"
+      pattern: "*.spectra-asm.hist"
+  - assembly_qv:
+      type: file
+      description: "Assembly consensus quality estimation"
+      pattern: "*.qv"
+  - scaffold_qv:
+      type: file
+      description: "Scaffold consensus quality estimation"
+      pattern: "*.qv"
+  - read_ploidy:
+      type: file
+      description: "Ploidy estimate from read k-mer database"
+      pattern: "*.hist.ploidy"
+
+authors:
+  - "@mahesh-panchal"

tests/config/pytest_modules.yml

@@ -1398,6 +1398,10 @@ meningotype:
   - modules/meningotype/**
   - tests/modules/meningotype/**
 
+merqury:
+  - modules/merqury/**
+  - tests/modules/merqury/**
+
 merquryfk/katcomp:
   - modules/merquryfk/katcomp/**
   - tests/modules/merquryfk/katcomp/**

tests/modules/merqury/main.nf

@@ -0,0 +1,26 @@
+#!/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'
+include { MERQURY        } from '../../../modules/merqury/main.nf'
+
+workflow test_merqury {
+
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        [
+            file(params.test_data['homo_sapiens']['illumina']['test_1_fastq_gz'], checkIfExists: true),
+            file(params.test_data['homo_sapiens']['illumina']['test_2_fastq_gz'], checkIfExists: true)
+        ]
+    ]
+    assembly = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+
+    MERYL_COUNT ( input )
+    MERYL_UNIONSUM ( MERYL_COUNT.out.meryl_db )
+    MERQURY ( MERYL_UNIONSUM.out.meryl_db.join( Channel.value( assembly ) ) )
+}

tests/modules/merqury/nextflow.config

@@ -0,0 +1,10 @@
+process {
+
+    withName: 'MERQURY' {
+        publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    }
+
+    withName: 'MERYL_COUNT' {
+        ext.args = 'k=21'
+    }
+}

tests/modules/merqury/test.yml

@@ -0,0 +1,34 @@
+- name: merqury test_merqury
+  command: nextflow run ./tests/modules/merqury -entry test_merqury -c ./tests/config/nextflow.config  -c ./tests/modules/merqury/nextflow.config
+  tags:
+    - merqury
+  files:
+    - path: output/merqury/genome_only.bed
+      md5sum: b611f22cde0e410a2ca07c1eefd042d3
+    - path: output/merqury/genome_only.wig
+      md5sum: 19cf44989af72af597ef80d3489b4882
+    - path: output/merqury/test.completeness.stats
+    - path: output/merqury/test.dist_only.hist
+      md5sum: e2e6b54b0febef1f0fcf24cd2afd0b7a
+    - path: output/merqury/test.genome.qv
+      md5sum: c554315aabcc4207c367805cf3090da3
+    - path: output/merqury/test.genome.spectra-cn.fl.png
+      md5sum: 3265701cbb1ddaed6d5cb6b4560564fd
+    - path: output/merqury/test.genome.spectra-cn.hist
+      md5sum: 6140a138ba47cb2b97814c93f80b2575
+    - path: output/merqury/test.genome.spectra-cn.ln.png
+      md5sum: 6386b604a8f0fbc43a1e3473ad9a779e
+    - path: output/merqury/test.genome.spectra-cn.st.png
+      md5sum: d5af80c91d23b182589b0ec131047f00
+    - path: output/merqury/test.qv
+      md5sum: 6e04952bc182221c8b9e242dc3298808
+    - path: output/merqury/test.spectra-asm.fl.png
+      md5sum: 60920481d988018a4f36a9be5b10c4ec
+    - path: output/merqury/test.spectra-asm.hist
+      md5sum: 541c9d1f87ab5c44df5e9e0acc440f8d
+    - path: output/merqury/test.spectra-asm.ln.png
+      md5sum: 54790dd54b5d8948d3676d32da1972df
+    - path: output/merqury/test.spectra-asm.st.png
+      md5sum: 010e013b5411ab2790e9a93ee70a574a
+    - path: output/merqury/test.unionsumdb.hist.ploidy
+      md5sum: f6904468b41a495c7ce255a7a5f3a302