Add CHECKM/QA (#2137)

* Bump checkM version and add support for pre-defined database

* Add markfile output to checkm/lineagewf and add checkm/qa

* Remove unnecessary BAM from tests

* Fix channels to ensure single meta

* Blah

* Use secondary name evwyhere

* Remove second incovcatiojn

* Don't run module directrly for join

* And for the other test

* Correct module name

* Fix name again

* Try and fix out_format 9 not being picked up

* Contains to matches to allow regex

* Fix tests (finally)

* Fix verisoning

* [automated] Fix linting with Prettier

* Update modules/checkm/qa/main.nf

Co-authored-by: nvnieuwk <101190534+nvnieuwk@users.noreply.github.com>

Co-authored-by: nf-core-bot <core@nf-co.re>
Co-authored-by: nvnieuwk <101190534+nvnieuwk@users.noreply.github.com>

modules/checkm/lineagewf/main.nf

@@ -13,9 +13,10 @@ process CHECKM_LINEAGEWF {
     path db
 
     output:
-    tuple val(meta), path("${prefix}")    , emit: checkm_output
-    tuple val(meta), path("${prefix}.tsv"), emit: checkm_tsv
-    path "versions.yml"                   , emit: versions
+    tuple val(meta), path("${prefix}")           , emit: checkm_output
+    tuple val(meta), path("${prefix}/lineage.ms"), emit: marker_file
+    tuple val(meta), path("${prefix}.tsv")       , emit: checkm_tsv
+    path "versions.yml"                          , emit: versions
 
     when:
     task.ext.when == null || task.ext.when

modules/checkm/lineagewf/meta.yml

@@ -52,6 +52,10 @@ output:
       type: directory
       description: CheckM output directory
       pattern: "*/"
+  - checkm_output:
+      type: file
+      description: Lineage markfer file
+      pattern: "lineage.ms"
   - checkm_tsv:
       type: file
       description: CheckM summary completeness statistics table

modules/checkm/qa/main.nf

@@ -0,0 +1,44 @@
+process CHECKM_QA {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::checkm-genome=1.2.1" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/checkm-genome:1.2.1--pyhdfd78af_0' :
+        'quay.io/biocontainers/checkm-genome:1.2.1--pyhdfd78af_0' }"
+
+    input:
+    tuple val(meta), path(analysis_dir), path(marker_file), path(coverage_file)
+    path exclude_marker_file
+
+    output:
+    tuple val(meta), path("${prefix}.txt")  , optional: true, emit: output
+    tuple val(meta), path("${prefix}.fasta"), optional: true, emit: fasta
+    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}"
+    suffix       = task.ext.args?.matches(".*-o 9.*|.*--out_file 9.*") ? "fasta" : "txt"
+    def coverage = coverage_file                                   ? "--coverage_file ${coverage_file}"  : ""
+    def exclude  = exclude_marker_file                             ? "--exclude_markers ${marker_filer}" : ""
+    """
+    checkm \\
+        qa \\
+        --threads ${task.cpus} \\
+        --file ${prefix}.${suffix} \\
+        $marker_file \\
+        $analysis_dir \\
+        $coverage \\
+        $exclude \\
+        $args
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        checkm: \$( checkm 2>&1 | grep '...:::' | sed 's/.*CheckM v//;s/ .*//' )
+    END_VERSIONS
+    """
+}

modules/checkm/qa/meta.yml

@@ -0,0 +1,67 @@
+name: checkm_qa
+description: CheckM provides a set of tools for assessing the quality of genomes recovered from isolates, single cells, or metagenomes.
+keywords:
+  - checkm
+  - mag
+  - metagenome
+  - quality
+  - isolates
+  - microbes
+  - single cells
+  - completeness
+  - contamination
+  - bins
+  - genome bins
+  - qa
+  - quality assurnce
+tools:
+  - checkm:
+      description: Assess the quality of microbial genomes recovered from isolates, single cells, and metagenomes.
+      homepage: https://ecogenomics.github.io/CheckM/
+      documentation: https://github.com/Ecogenomics/CheckM/wiki
+      tool_dev_url: https://github.com/Ecogenomics/CheckM
+      doi: "10.1101/gr.186072.114"
+      licence: ["GPL v3"]
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - analysis_dir:
+      type: file
+      description: Directory containing output of checkm/analyze or checkm/lineage_wf etc.
+      pattern: "*"
+  - marker_file:
+      type: file
+      description: Marker file specified during checkm/analyze or produced by checkm/{lineage,taxonomy}_wf
+      pattern: "*.ms"
+  - coverage_file:
+      type: file
+      description: File containing coverage of each sequence (generated by checkm coverage)
+  - exclude_marker_file:
+      type: file
+      description: File specifying markers to exclude from marker sets
+
+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"
+  - output:
+      type: file
+      description: "Default completeness statistics in various formats, as specified with --out_format (excluding option: 9)"
+      pattern: "*.txt"
+  - fasta:
+      type: file
+      description: Output in fasta format (only if --out_format 9)
+      pattern: "*.fasta"
+
+authors:
+  - "@jfy133"

tests/config/pytest_modules.yml

@@ -499,6 +499,10 @@ checkm/lineagewf:
   - modules/checkm/lineagewf/**
   - tests/modules/checkm/lineagewf/**
 
+checkm/qa:
+  - modules/checkm/qa/**
+  - tests/modules/checkm/qa/**
+
 chromap/chromap:
   - modules/chromap/chromap/**
   - tests/modules/chromap/chromap/**

tests/modules/checkm/qa/main.nf

@@ -0,0 +1,44 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { CHECKM_LINEAGEWF              } from '../../../../modules/checkm/lineagewf/main.nf'
+include { CHECKM_LINEAGEWF as CHECKM_LINEAGEWF_FASTA        } from '../../../../modules/checkm/lineagewf/main.nf'
+include { CHECKM_QA                     } from '../../../../modules/checkm/qa/main.nf'
+include { CHECKM_QA as CHECKM_QA_FASTA  } from '../../../../modules/checkm/qa/main.nf'
+
+workflow test_checkm_qa {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['illumina']['contigs_fasta'], checkIfExists: true) ]
+    fasta_ext = 'fasta'
+
+    CHECKM_LINEAGEWF ( input, fasta_ext, [] )
+
+    ch_checkmqa_input = CHECKM_LINEAGEWF.out.checkm_output
+        .join(CHECKM_LINEAGEWF.out.marker_file)
+        .map{
+            meta, dir, marker ->
+            [ meta, dir, marker, []]
+        }
+
+    CHECKM_QA ( ch_checkmqa_input, [] )
+}
+
+workflow test_checkm_qa_fasta {
+
+    input = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['sarscov2']['illumina']['contigs_fasta'], checkIfExists: true) ]
+    fasta_ext = 'fasta'
+
+    CHECKM_LINEAGEWF_FASTA ( input, fasta_ext, [] )
+
+    ch_checkmqa_input =  CHECKM_LINEAGEWF_FASTA.out.checkm_output
+        .join(CHECKM_LINEAGEWF_FASTA.out.marker_file)
+        .map{
+            meta, dir, marker ->
+            [ meta, dir, marker, []]
+        }
+
+    CHECKM_QA_FASTA ( ch_checkmqa_input, [] )
+}

tests/modules/checkm/qa/nextflow.config

@@ -0,0 +1,16 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+    withName: CHECKM_QA {
+        ext.prefix = { "${meta.id}.qa" }
+    }
+
+
+    withName: CHECKM_QA_FASTA {
+        ext.args = "-o 9"
+        ext.prefix = { "${meta.id}.qa" }
+    }
+
+
+}

tests/modules/checkm/qa/test.yml

@@ -0,0 +1,16 @@
+- name: checkm qa test_checkm_qa
+  command: nextflow run ./tests/modules/checkm/qa -entry test_checkm_qa -c ./tests/config/nextflow.config  -c ./tests/modules/checkm/qa/nextflow.config
+  tags:
+    - checkm/qa
+    - checkm
+  files:
+    - path: output/checkm/test.qa.txt
+      md5sum: 645f4282569afb4b171396732b2d2582
+
+- name: checkm qa test_checkm_qa_fasta
+  command: nextflow run ./tests/modules/checkm/qa -entry test_checkm_qa_fasta -c ./tests/config/nextflow.config  -c ./tests/modules/checkm/qa/nextflow.config
+  tags:
+    - checkm/qa
+    - checkm
+  files:
+    - path: output/checkm/test.qa.fasta