Add MerquryFK (#1778)

* Add MerquryFK

* Change resource allocation

modules/merquryfk/merquryfk/main.nf

@@ -0,0 +1,55 @@
+process MERQURYFK_MERQURYFK {
+    tag "$meta.id"
+    label 'process_medium'
+
+    if (params.enable_conda) {
+        error "Conda environments cannot be used when using the FastK tool. Please use docker or singularity containers."
+    }
+    container 'ghcr.io/nbisweden/fastk_genescopefk_merquryfk:1.0'
+
+    input:
+    tuple val(meta), path(fastk_hist), path(fastk_ktab), path(assembly)
+
+    output:
+    tuple val(meta), path("${prefix}.completeness.stats") , emit: stats
+    tuple val(meta), path("${prefix}.*_only.bed")         , emit: bed
+    tuple val(meta), path("${prefix}.*.qv")               , emit: assembly_qv
+    tuple val(meta), path("${prefix}.*.spectra-cn.fl.png"), emit: spectra_cn_fl_png,  optional: true
+    tuple val(meta), path("${prefix}.*.spectra-cn.fl.pdf"), emit: spectra_cn_fl_pdf,  optional: true
+    tuple val(meta), path("${prefix}.*.spectra-cn.ln.png"), emit: spectra_cn_ln_png,  optional: true
+    tuple val(meta), path("${prefix}.*.spectra-cn.ln.pdf"), emit: spectra_cn_ln_pdf,  optional: true
+    tuple val(meta), path("${prefix}.*.spectra-cn.st.png"), emit: spectra_cn_st_png,  optional: true
+    tuple val(meta), path("${prefix}.*.spectra-cn.st.pdf"), emit: spectra_cn_st_pdf,  optional: true
+    tuple val(meta), path("${prefix}.qv")                 , emit: qv
+    tuple val(meta), path("${prefix}.spectra-asm.fl.png") , emit: spectra_asm_fl_png, optional: true
+    tuple val(meta), path("${prefix}.spectra-asm.fl.pdf") , emit: spectra_asm_fl_pdf, optional: true
+    tuple val(meta), path("${prefix}.spectra-asm.ln.png") , emit: spectra_asm_ln_png, optional: true
+    tuple val(meta), path("${prefix}.spectra-asm.ln.pdf") , emit: spectra_asm_ln_pdf, optional: true
+    tuple val(meta), path("${prefix}.spectra-asm.st.png") , emit: spectra_asm_st_png, optional: true
+    tuple val(meta), path("${prefix}.spectra-asm.st.pdf") , emit: spectra_asm_st_pdf, optional: true
+    path "versions.yml"                                   , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    prefix = task.ext.prefix ?: "${meta.id}"
+    def FASTK_VERSION = 'f18a4e6d2207539f7b84461daebc54530a9559b0'
+    def MERQURY_VERSION = '8f3ab706e4cf4d7b7d1dfe5739859e3ebd26c494'
+    """
+    MerquryFK \\
+        $args \\
+        -T$task.cpus \\
+        ${fastk_ktab.find{ it.toString().endsWith(".ktab") }} \\
+        $assembly \\
+        $prefix
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        fastk: $FASTK_VERSION
+        merquryfk: $MERQURY_VERSION
+        r: \$( R --version | sed '1!d; s/.*version //; s/ .*//' )
+    END_VERSIONS
+    """
+}

modules/merquryfk/merquryfk/meta.yml

@@ -0,0 +1,109 @@
+name: "merquryfk_merquryfk"
+description: FastK based version of Merqury
+keywords:
+  - sort
+tools:
+  - "merquryfk":
+      description: "FastK based version of Merqury"
+      homepage: "https://github.com/thegenemyers/MERQURY.FK"
+      documentation: ""
+      tool_dev_url: "https://github.com/thegenemyers/MERQURY.FK"
+      doi: ""
+      licence: "https://github.com/thegenemyers/MERQURY.FK/blob/main/LICENSE"
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - fastk_hist:
+      type: file
+      description: A histogram files from the program FastK
+      pattern: "*.hist"
+  - fastk_ktab:
+      type: file
+      description: Histogram ktab files from the program FastK (option -t)
+      pattern: "*.ktab*"
+  - assembly:
+      type: file
+      description: Genome assembly files (fasta format)
+      pattern: ".fasta"
+
+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"
+  - stats:
+      type: file
+      description: Assembly statistics file
+      pattern: "*.completeness.stats"
+  - bed:
+      type: file
+      description: Assembly only kmer positions not supported by reads in bed format
+      pattern: "*_only.bed"
+  - 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_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_cn_fl_pdf:
+      type: file
+      description: "Unstacked copy number spectra filled plot in PDF format"
+      pattern: "*.spectra-cn.fl.pdf"
+  - spectra_cn_ln_pdf:
+      type: file
+      description: "Unstacked copy number spectra line plot in PDF format"
+      pattern: "*.spectra-cn.ln.pdf"
+  - spectra_cn_st_pdf:
+      type: file
+      description: "Stacked copy number spectra line plot in PDF format"
+      pattern: "*.spectra-cn.st.pdf"
+  - spectra_asm_fl_pdf:
+      type: file
+      description: "Unstacked assembly spectra filled plot in PDF format"
+      pattern: "*.spectra-asm.fl.pdf"
+  - spectra_asm_ln_pdf:
+      type: file
+      description: "Unstacked assembly spectra line plot in PDF format"
+      pattern: "*.spectra-asm.ln.pdf"
+  - spectra_asm_st_pdf:
+      type: file
+      description: "Stacked assembly spectra line plot in PDF format"
+      pattern: "*.spectra-asm.st.pdf"
+  - assembly_qv:
+      type: file
+      description: "error and qv table for each scaffold of the assembly"
+      pattern: "*.qv"
+  - qv:
+      type: file
+      description: "error and qv of each assembly as a whole"
+      pattern: "*.qv"
+
+authors:
+  - "@mahesh-panchal"

tests/config/pytest_modules.yml

@@ -1330,6 +1330,10 @@ meningotype:
   - modules/meningotype/**
   - tests/modules/meningotype/**
 
+merquryfk/merquryfk:
+  - modules/merquryfk/merquryfk/**
+  - tests/modules/merquryfk/merquryfk/**
+
 meryl/count:
   - modules/meryl/count/**
   - tests/modules/meryl/count/**

tests/modules/merquryfk/merquryfk/main.nf

@@ -0,0 +1,42 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { FASTK_FASTK         } from '../../../../modules/fastk/fastk/main.nf'
+include { MERQURYFK_MERQURYFK } from '../../../../modules/merquryfk/merquryfk/main.nf'
+
+workflow test_merquryfk_merquryfk_png {
+
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+    ]
+    assembly = [
+        [ id:'test', single_end:true ], // meta map
+        file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+
+    FASTK_FASTK ( input )
+    MERQURYFK_MERQURYFK ( FASTK_FASTK.out.hist
+        .join( FASTK_FASTK.out.ktab )
+        .join( Channel.value( assembly ) )
+    )
+}
+
+workflow test_merquryfk_merquryfk_pdf {
+
+    input = [
+        [ id:'test', single_end:true ], // meta map
+        file(params.test_data['homo_sapiens']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+    ]
+    assembly = [
+        [ id:'test', single_end:true ], // meta map
+        file(params.test_data['homo_sapiens']['genome']['genome_fasta'], checkIfExists: true)
+    ]
+
+    FASTK_FASTK ( input )
+    MERQURYFK_MERQURYFK ( FASTK_FASTK.out.hist
+        .join( FASTK_FASTK.out.ktab )
+        .join( Channel.value( assembly ) )
+    )
+}

tests/modules/merquryfk/merquryfk/nextflow.config

@@ -0,0 +1,16 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName: 'FASTK_.*' {
+        ext.args   = '-t'
+        publishDir = [ enabled: false ]
+    }
+    withName: 'test_merquryfk_merquryfk_png:MERQURYFK_MERQURYFK' {
+        ext.args   = '-lfs'
+    }
+    withName: 'test_merquryfk_merquryfk_pdf:MERQURYFK_MERQURYFK' {
+        ext.args   = '-lfs -pdf'
+    }
+
+}

tests/modules/merquryfk/merquryfk/test.yml

@@ -0,0 +1,47 @@
+- name: merquryfk merquryfk test_merquryfk_merquryfk_png
+  command: nextflow run ./tests/modules/merquryfk/merquryfk -entry test_merquryfk_merquryfk_png -c ./tests/config/nextflow.config  -c ./tests/modules/merquryfk/merquryfk/nextflow.config
+  tags:
+    - merquryfk/merquryfk
+    - merquryfk
+  files:
+    - path: output/merquryfk/test.completeness.stats
+      md5sum: 797224fa75606bbda7f62caae7c9151c
+    - path: output/merquryfk/test.genome.qv
+      md5sum: 1e11be8abfa2230024042832f58f96f9
+    - path: output/merquryfk/test.genome.spectra-cn.fl.png
+      md5sum: 50bad6b85b8f80b8411b50e4119bc959
+    - path: output/merquryfk/test.genome.spectra-cn.ln.png
+      md5sum: 50bad6b85b8f80b8411b50e4119bc959
+    - path: output/merquryfk/test.genome.spectra-cn.st.png
+      md5sum: 50bad6b85b8f80b8411b50e4119bc959
+    - path: output/merquryfk/test.genome_only.bed
+      md5sum: 662d73e8a0019708feb538ec53c220f7
+    - path: output/merquryfk/test.qv
+      md5sum: 02b4d5c639ca706ff707bad89e29d90e
+    - path: output/merquryfk/test.spectra-asm.fl.png
+      md5sum: bb9ac38fe8991fbacbe791e562626d89
+    - path: output/merquryfk/test.spectra-asm.ln.png
+      md5sum: 48534a9d5e2ce9365eb1f0397c76f337
+    - path: output/merquryfk/test.spectra-asm.st.png
+      md5sum: 5e031a8b9ba2800e291721a11f4d9c00
+
+- name: merquryfk merquryfk test_merquryfk_merquryfk_pdf
+  command: nextflow run ./tests/modules/merquryfk/merquryfk -entry test_merquryfk_merquryfk_pdf -c ./tests/config/nextflow.config  -c ./tests/modules/merquryfk/merquryfk/nextflow.config
+  tags:
+    - merquryfk/merquryfk
+    - merquryfk
+  files:
+    - path: output/merquryfk/test.completeness.stats
+      md5sum: 797224fa75606bbda7f62caae7c9151c
+    - path: output/merquryfk/test.genome.qv
+      md5sum: 1e11be8abfa2230024042832f58f96f9
+    - path: output/merquryfk/test.genome.spectra-cn.fl.pdf
+    - path: output/merquryfk/test.genome.spectra-cn.ln.pdf
+    - path: output/merquryfk/test.genome.spectra-cn.st.pdf
+    - path: output/merquryfk/test.genome_only.bed
+      md5sum: 662d73e8a0019708feb538ec53c220f7
+    - path: output/merquryfk/test.qv
+      md5sum: 02b4d5c639ca706ff707bad89e29d90e
+    - path: output/merquryfk/test.spectra-asm.fl.pdf
+    - path: output/merquryfk/test.spectra-asm.ln.pdf
+    - path: output/merquryfk/test.spectra-asm.st.pdf