add FCS adaptor (#2033)

* add FCS adaptor

* run prettier

* fix EClint

* add keywords to meta

* fix docker

Co-authored-by: Simon Pearce <24893913+SPPearce@users.noreply.github.com>

modules/fcs/fcsadaptor/main.nf

@@ -0,0 +1,49 @@
+process FCS_FCSADAPTOR {
+    tag "$meta.id"
+    label 'process_low'
+
+    if (params.enable_conda) {
+        exit 1, "Conda environments cannot be used when using the FCS tool. Please use docker or singularity containers."
+    }
+    // WARN: Version information not provided by tool on CLI. Please update version string below when bumping container versions.
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://ftp.ncbi.nlm.nih.gov/genomes/TOOLS/FCS/releases/0.2.3/fcs-adaptor.0.2.3.sif':
+        'ncbi/fcs-adaptor:0.2.3' }"
+
+    input:
+    tuple val(meta), path(assembly)
+
+    output:
+    tuple val(meta), path("*.cleaned_sequences.fa.gz"), emit: cleaned_assembly
+    tuple val(meta), path("*.fcs_adaptor_report.txt") , emit: adaptor_report
+    tuple val(meta), path("*.fcs_adaptor.log")        , emit: log
+    tuple val(meta), path("*.pipeline_args.yaml")     , emit: pipeline_args
+    tuple val(meta), path("*.skipped_trims.jsonl")    , emit: skipped_trims
+    path "versions.yml"                               , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: '--prok' // --prok || --euk
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def FCSADAPTOR_VERSION = '0.2.3' // WARN: Version information not provided by tool on CLI. Please update this string when bumping container versions.
+    """
+    /app/fcs/bin/av_screen_x \\
+        -o output/ \\
+        $args \\
+        $assembly
+
+    # compress and/or rename files with prefix
+    gzip -cf output/cleaned_sequences/* > "${prefix}.cleaned_sequences.fa.gz"
+    cp "output/fcs_adaptor_report.txt"    "${prefix}.fcs_adaptor_report.txt"
+    cp "output/fcs_adaptor.log"           "${prefix}.fcs_adaptor.log"
+    cp "output/pipeline_args.yaml"        "${prefix}.pipeline_args.yaml"
+    cp "output/skipped_trims.jsonl"       "${prefix}.skipped_trims.jsonl"
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        FCS-adaptor: $FCSADAPTOR_VERSION
+    END_VERSIONS
+    """
+}

modules/fcs/fcsadaptor/meta.yml

@@ -0,0 +1,62 @@
+name: "fcs_fcsadaptor"
+description: Run NCBI's FCS adaptor on assembled genomes
+keywords:
+  - assembly
+  - genomics
+  - quality control
+  - contamination
+  - NCBI
+tools:
+  - "fcs":
+      description: |
+        The Foreign Contamination Screening (FCS) tool rapidly detects contaminants from foreign
+        organisms in genome assemblies to prepare your data for submission. Therefore, the
+        submission process to NCBI is faster and fewer contaminated genomes are submitted.
+        This reduces errors in analyses and conclusions, not just for the original data submitter
+        but for all subsequent users of the assembly.
+      homepage: "https://www.ncbi.nlm.nih.gov/data-hub/cgr/data-quality-tools/"
+      documentation: "https://github.com/ncbi/fcs/wiki/FCS-adaptor"
+      tool_dev_url: "https://github.com/ncbi/fcs"
+      doi: ""
+      licence: "United States Government Work"
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - assembly:
+      type: file
+      description: assembly fasta 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"
+  - cleaned_assembly:
+      type: file
+      description: Cleaned assembly in fasta format
+      pattern: "*.{cleaned_sequences.fa.gz}"
+  - adaptor_report:
+      type: file
+      description: Report of identified adaptors
+      pattern: "*.{fcs_adaptor_report.txt}"
+  - log:
+      type: file
+      description: Log file
+      pattern: "*.{fcs_adaptor.log}"
+  - pipeline_args:
+      type: file
+      description: Run arguments
+      pattern: "*.{pipeline_args.yaml}"
+  - skipped_trims:
+      type: file
+      description: Skipped trim information
+      pattern: "*.{skipped_trims.jsonl}"
+authors:
+  - "@d4straub"

tests/config/pytest_modules.yml

@@ -803,6 +803,10 @@ fasttree:
   - modules/fasttree/**
   - tests/modules/fasttree/**
 
+fcs/fcsadaptor:
+  - modules/fcs/fcsadaptor/**
+  - tests/modules/fcs/fcsadaptor/**
+
 ffq:
   - modules/ffq/**
   - tests/modules/ffq/**

tests/modules/fcs/fcsadaptor/main.nf

@@ -0,0 +1,15 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { FCS_FCSADAPTOR } from '../../../../modules/fcs/fcsadaptor/main.nf'
+
+workflow test_fcs_fcsadaptor {
+    
+    input = [
+        [ id:'test', single_end:false ], // meta map
+        file(params.test_data['bacteroides_fragilis']['genome']['genome_fna_gz'], checkIfExists: true)
+    ]
+
+    FCS_FCSADAPTOR ( input )
+}

tests/modules/fcs/fcsadaptor/nextflow.config

@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+    
+}

tests/modules/fcs/fcsadaptor/test.yml

@@ -0,0 +1,15 @@
+- name: fcs fcsadaptor test_fcs_fcsadaptor
+  command: nextflow run ./tests/modules/fcs/fcsadaptor -entry test_fcs_fcsadaptor -c ./tests/config/nextflow.config  -c ./tests/modules/fcs/fcsadaptor/nextflow.config
+  tags:
+    - fcs/fcsadaptor
+    - fcs
+  files:
+    - path: output/fcs/test.cleaned_sequences.fa.gz
+      md5sum: 2f331e59b352418420a35236091bf93c
+    - path: output/fcs/test.fcs_adaptor.log
+      contains: ["Original command:", "Executing:", "INFO:cwltool:[workflow ] completed success"]
+    - path: output/fcs/test.fcs_adaptor_report.txt
+      md5sum: 27375be4671e01d2d2674ddd1d44414a
+    - path: output/fcs/test.pipeline_args.yaml
+      md5sum: 80effed2a75550a6a861c47f4c30449d
+    - path: output/fcs/test.skipped_trims.jsonl