Update DASTool to 1.1.4 (#1471)

* fix: remove left-over unnecessary code

* Update DASTool

* Fix tests

* Fix test.ymls

* Fix container build version

* Make tests less strict to account for variability

* Apply suggestions from code review

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

* Add missing description

Co-authored-by: Sateesh Peri <33637490+sateeshperi@users.noreply.github.com>
Co-authored-by: Daniel Straub <42973691+d4straub@users.noreply.github.com>

modules/dastool/dastool/main.nf

@@ -2,27 +2,28 @@ process DASTOOL_DASTOOL {
     tag "$meta.id"
     label 'process_medium'
 
-    conda (params.enable_conda ? "bioconda::das_tool=1.1.3" : null)
+    conda (params.enable_conda ? "bioconda::das_tool=1.1.4" : null)
     container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
-        'https://depot.galaxyproject.org/singularity/das_tool:1.1.3--r41hdfd78af_0' :
+        'https://depot.galaxyproject.org/singularity/das_tool:1.1.4--r41hdfd78af_1' :
-        'quay.io/biocontainers/das_tool:1.1.3--r41hdfd78af_0' }"
+        'quay.io/biocontainers/das_tool:1.1.4--r41hdfd78af_1' }"
 
     input:
     tuple val(meta), path(contigs), path(bins)
     path(proteins)
     path(db_directory)
-    val(search_engine)
 
     output:
     tuple val(meta), path("*.log")                                      , emit: log
-    tuple val(meta), path("*_summary.txt")                              , emit: summary
+    tuple val(meta), path("*_summary.tsv")                              , emit: summary
-    tuple val(meta), path("*_DASTool_scaffolds2bin.txt")                , emit: scaffolds2bin
+    tuple val(meta), path("*_DASTool_contig2bin.tsv")                   , emit: contig2bin
     tuple val(meta), path("*.eval")                     , optional: true, emit: eval
     tuple val(meta), path("*_DASTool_bins/*.fa")        , optional: true, emit: bins
     tuple val(meta), path("*.pdf")                      , optional: true, emit: pdfs
-    tuple val(meta), path("*.proteins.faa")             , optional: true, emit: fasta_proteins
+    tuple val(meta), path("*.candidates.faa")           , optional: true, emit: fasta_proteins
+    tuple val(meta), path("*.faa")                      , optional: true, emit: candidates_faa
     tuple val(meta), path("*.archaea.scg")              , optional: true, emit: fasta_archaea_scg
     tuple val(meta), path("*.bacteria.scg")             , optional: true, emit: fasta_bacteria_scg
+    tuple val(meta), path("*.b6")                       , optional: true, emit: b6
     tuple val(meta), path("*.seqlength")                , optional: true, emit: seqlength
     path "versions.yml"                                 , emit: versions
 
@@ -33,17 +34,12 @@ process DASTOOL_DASTOOL {
     def args = task.ext.args ?: ''
     def prefix = task.ext.prefix ?: "${meta.id}"
     def bin_list = bins instanceof List ? bins.join(",") : "$bins"
-    def engine = search_engine ? "--search_engine $search_engine" : "--search_engine diamond"
     def db_dir = db_directory ? "--db_directory $db_directory" : ""
     def clean_contigs = contigs.toString() - ".gz"
     def decompress_contigs = contigs.toString() == clean_contigs ? "" : "gunzip -q -f $contigs"
-    def decompress_proteins = proteins ? "gunzip -f $proteins" : ""
     def clean_proteins = proteins ? proteins.toString() - ".gz" : ""
-    def proteins_pred = proteins ? "--proteins $clean_proteins" : ""
+    def decompress_proteins = proteins ? "gunzip -f $proteins" : ""
-
+    def proteins_pred = proteins ? "-p $clean_proteins" : ""
-    if (! search_engine) {
-        log.info('[DAS_Tool] Default search engine (USEARCH) is proprietary software and not available in bioconda. Using DIAMOND as alternative.')
-    }
 
     """
     $decompress_proteins
@@ -53,15 +49,14 @@ process DASTOOL_DASTOOL {
         $args \\
         $proteins_pred \\
         $db_dir \\
-        $engine \\
         -t $task.cpus \\
-        --bins $bin_list \\
+        -i $bin_list \\
         -c $clean_contigs \\
         -o $prefix
 
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        dastool: \$( DAS_Tool --version 2>&1 | grep "DAS Tool" | sed 's/DAS Tool version //' )
+        dastool: \$( DAS_Tool --version 2>&1 | grep "DAS Tool" | sed 's/DAS Tool //' )
     END_VERSIONS
     """
 }

modules/dastool/dastool/meta.yml

@@ -34,8 +34,8 @@ input:
       pattern: "*.{fa.gz,fas.gz,fasta.gz}"
   - bins:
       type: file
-      description: "Scaffolds2bin tabular file generated with dastool/scaffolds2bin"
+      description: "FastaToContig2Bin tabular file generated with dastool/fastatocontig2bin"
-      pattern: "*.scaffolds2bin.tsv"
+      pattern: "*.tsv"
   - proteins:
       type: file
       description: Predicted proteins in prodigal fasta format (>scaffoldID_geneNo)
@@ -43,9 +43,6 @@ input:
   - db_directory:
       type: file
       description: (optional) Directory of single copy gene database.
-  - search_engine:
-      type: val
-      description: Engine used for single copy gene identification. USEARCH is not supported due to it being proprietary [blast/diamond]
 
 output:
   - meta:
@@ -65,14 +62,17 @@ output:
       type: file
       description: Summary of output bins including quality and completeness estimates
       pattern: "*summary.txt"
-  - scaffolds2bin:
+  - contig2bin:
       type: file
       description: Scaffolds to bin file of output bins
-      pattern: "*.scaffolds2bin.txt"
+      pattern: "*.contig2bin.txt"
   - eval:
       type: file
       description: Quality and completeness estimates of input bin sets
       pattern: "*.eval"
+  - bins:
+      description: Final refined bins in fasta format
+      pattern: "*.fa"
   - pdfs:
       type: file
       description: Plots showing the amount of high quality bins and score distribution of bins per method
@@ -89,6 +89,10 @@ output:
       type: file
       description: Results of bacterial single-copy-gene prediction
       pattern: "*.bacteria.scg"
+  - b6:
+      type: file
+      description: Results in b6 format
+      pattern: "*.b6"
   - seqlength:
       type: file
       description: Summary of contig lengths

modules/dastool/fastatocontig2bin/main.nf

@@ -0,0 +1,41 @@
+process DASTOOL_FASTATOCONTIG2BIN {
+    tag "$meta.id"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::das_tool=1.1.4" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/das_tool:1.1.4--r41hdfd78af_1' :
+        'quay.io/biocontainers/das_tool:1.1.4--r41hdfd78af_1' }"
+
+    input:
+    tuple val(meta), path(fasta)
+    val(extension)
+
+    output:
+    tuple val(meta), path("*.tsv"), emit: fastatocontig2bin
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    def file_extension = extension ? extension : "fasta"
+    def clean_fasta = fasta.toString() - ".gz"
+    def decompress_fasta = fasta.toString() == clean_fasta ? "" : "gunzip -q -f $fasta"
+    """
+    $decompress_fasta
+
+    Fasta_to_Contig2Bin.sh \\
+        $args \\
+        -i . \\
+        -e $file_extension \\
+        > ${prefix}.tsv
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        dastool: \$( DAS_Tool --version 2>&1 | grep "DAS Tool" | sed 's/DAS Tool //' )
+    END_VERSIONS
+    """
+}

modules/dastool/fastatocontig2bin/meta.yml

@@ -0,0 +1,56 @@
+name: dastool_fastatocontig2bin
+description: Helper script to convert a set of bins in fasta format to tabular scaffolds2bin format
+keywords:
+  - binning
+  - das tool
+  - table
+  - de novo
+  - bins
+  - contigs
+  - assembly
+  - das_tool
+tools:
+  - dastool:
+      description: |
+        DAS Tool is an automated method that integrates the results
+        of a flexible number of binning algorithms to calculate an optimized, non-redundant
+        set of bins from a single assembly.
+
+      homepage: https://github.com/cmks/DAS_Tool
+      documentation: https://github.com/cmks/DAS_Tool
+      tool_dev_url: https://github.com/cmks/DAS_Tool
+      doi: "10.1038/s41564-018-0171-1"
+      licence: ["BSD"]
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+  - fasta:
+      type: file
+      description: Fasta of list of fasta files recommended to be gathered via with .collect() of bins
+      pattern: "*.{fa,fa.gz,fas,fas.gz,fna,fna.gz,fasta,fasta.gz}"
+  - extension:
+      type: val
+      description: Fasta file extension (fa | fas | fasta | ...), without .gz suffix, if gzipped input.
+
+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"
+  - fastatocontig2bin:
+      type: file
+      description: tabular contig2bin file for DAS tool input
+      pattern: "*.tsv"
+
+authors:
+  - "@maxibor"
+  - "@jfy133"

tests/config/pytest_modules.yml

@@ -487,6 +487,10 @@ dastool/dastool:
   - modules/dastool/dastool/**
   - tests/modules/dastool/dastool/**
 
+dastool/fastatocontig2bin:
+  - modules/dastool/fastatocontig2bin/**
+  - tests/modules/dastool/fastatocontig2bin/**
+
 dastool/scaffolds2bin:
   - modules/dastool/scaffolds2bin/**
   - tests/modules/dastool/scaffolds2bin/**

tests/modules/dastool/dastool/main.nf

@@ -3,7 +3,7 @@ nextflow.enable.dsl = 2
 
 include { METABAT2_METABAT2 } from '../../../../modules/metabat2/metabat2/main.nf'
 include { METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS } from '../../../../modules/metabat2/jgisummarizebamcontigdepths/main.nf'
-include { DASTOOL_SCAFFOLDS2BIN } from '../../../../modules/dastool/scaffolds2bin/main.nf'
+include { DASTOOL_FASTATOCONTIG2BIN } from '../../../../modules/dastool/fastatocontig2bin/main.nf'
 include { DASTOOL_DASTOOL } from '../../../../modules/dastool/dastool/main.nf'
 
 workflow test_dastool_dastool {
@@ -21,13 +21,13 @@ workflow test_dastool_dastool {
 
     METABAT2_METABAT2 ( input_metabat2 )
 
-    DASTOOL_SCAFFOLDS2BIN ( METABAT2_METABAT2.out.fasta.collect(), "fa")
+    DASTOOL_FASTATOCONTIG2BIN ( METABAT2_METABAT2.out.fasta.collect(), "fa")
 
     Channel.of([ [ id:'test', single_end:false ], // meta map
                       file(params.test_data['bacteroides_fragilis']['genome']['genome_fna_gz'], checkIfExists: true)])
-        .join(DASTOOL_SCAFFOLDS2BIN.out.scaffolds2bin)
+        .join( DASTOOL_FASTATOCONTIG2BIN.out.fastatocontig2bin )
         .set {input_dastool}
-    
 
-    DASTOOL_DASTOOL ( input_dastool, [], [], [] )
+
+    DASTOOL_DASTOOL ( input_dastool, [], [] )
 }

tests/modules/dastool/dastool/test.yml

@@ -1,29 +1,28 @@
 - name: dastool dastool test_dastool_dastool
-  command: nextflow run ./tests/modules/dastool/dastool -entry test_dastool_dastool -c ./tests/config/nextflow.config -c ./tests/modules/dastool/dastool/nextflow.config
+  command: nextflow run tests/modules/dastool/dastool -entry test_dastool_dastool -c tests/config/nextflow.config
   tags:
-    - dastool
     - dastool/dastool
+    - dastool
   files:
     - path: output/dastool/test.seqlength
       md5sum: b815a5811008c36808a59b1d0dcfab24
     - path: output/dastool/test.tsv
       md5sum: 6e46c0be14dded7cb13af38f54feea47
     - path: output/dastool/test_DASTool.log
-      contains:
+    - path: output/dastool/test_DASTool_contig2bin.tsv
-        - "DAS Tool run on"
-    - path: output/dastool/test_DASTool_scaffolds2bin.txt
       md5sum: 6e46c0be14dded7cb13af38f54feea47
-    - path: output/dastool/test_DASTool_summary.txt
+    - path: output/dastool/test_DASTool_summary.tsv
-      md5sum: a3efa8717b30dfada78dc5ae9a3dc396
+      md5sum: ab9dd3709a59a69bc66030b9e0ff3d5b
+    - path: output/dastool/test_proteins.faa
+    - path: output/dastool/test_proteins.faa.all.b6
+      md5sum: 39c11237ef22ac73109aaac267e185d0
     - path: output/dastool/test_proteins.faa.archaea.scg
       md5sum: e79d82eecee25821d1658ea4f082601d
     - path: output/dastool/test_proteins.faa.bacteria.scg
       md5sum: 8132cfb17cf398d41c036ead55c96ffe
-    - path: output/dastool/test_test.tsv.eval
+    - path: output/dastool/test_proteins.faa.findSCG.b6
-      md5sum: a3efa8717b30dfada78dc5ae9a3dc396
+      md5sum: 48e90e12cd6c88d00608777dbc48a82a
-    - path: output/metabat2/bins/test.1.fa.gz
+    - path: output/dastool/test_proteins.faa.scg.candidates.faa
-      md5sum: 2b297bf557cc3831b800348859331268
+      md5sum: d94b7bed0f8aa9cf2824d72c548c537c
-    - path: output/metabat2/test.tsv.gz
+    - path: output/dastool/versions.yml
-      md5sum: 619338fa5019e361d5545ce385a6961f
+      md5sum: 004e04c6a38652df2e0c59c44e29c9de
-    - path: output/metabat2/test.txt.gz
-      md5sum: 745a0446af6ef68b930975e9ce5a95d6

tests/modules/dastool/fastatocontig2bin/main.nf

@@ -0,0 +1,48 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { GUNZIP                                } from '../../../../modules/gunzip/main.nf'
+include { METABAT2_METABAT2                     } from '../../../../modules/metabat2/metabat2/main.nf'
+include { METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS  } from '../../../../modules/metabat2/jgisummarizebamcontigdepths/main.nf'
+include { DASTOOL_FASTATOCONTIG2BIN                 } from '../../../../modules/dastool/fastatocontig2bin/main.nf'
+
+workflow test_dastool_fastatocontig2bin {
+
+    input_depth = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam'], checkIfExists: true),
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+
+    METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS ( input_depth )
+
+    Channel.fromPath(params.test_data['bacteroides_fragilis']['genome']['genome_fna_gz'], checkIfExists: true)
+        .map { it -> [[ id:'test', single_end:false ], it] }
+        .join(METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS.out.depth)
+        .set { input_metabat2 }
+
+    METABAT2_METABAT2 ( input_metabat2 )
+
+    DASTOOL_FASTATOCONTIG2BIN ( METABAT2_METABAT2.out.fasta.collect(), "fa")
+}
+
+workflow test_dastool_fastatocontig2bin_ungzipped {
+
+    input_depth = [ [ id:'test', single_end:false ], // meta map
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam'], checkIfExists: true),
+              file(params.test_data['bacteroides_fragilis']['illumina']['test1_paired_end_sorted_bam_bai'], checkIfExists: true) ]
+
+
+    METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS ( input_depth )
+
+    Channel.fromPath(params.test_data['bacteroides_fragilis']['genome']['genome_fna_gz'], checkIfExists: true)
+        .map { it -> [[ id:'test', single_end:false ], it] }
+        .join(METABAT2_JGISUMMARIZEBAMCONTIGDEPTHS.out.depth)
+        .set { input_metabat2 }
+
+    METABAT2_METABAT2 ( input_metabat2 )
+
+    // TODO test unzipped input files
+    ch_input_2_fastatocontig2bin = GUNZIP( METABAT2_METABAT2.out.fasta  ).gunzip
+
+    DASTOOL_FASTATOCONTIG2BIN ( ch_input_2_fastatocontig2bin, "fa")
+}

tests/modules/dastool/fastatocontig2bin/nextflow.config

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

tests/modules/dastool/fastatocontig2bin/test.yml

@@ -0,0 +1,20 @@
+- name: dastool fastatocontig2bin test_dastool_fastatocontig2bin
+  command: nextflow run tests/modules/dastool/fastatocontig2bin -entry test_dastool_fastatocontig2bin -c tests/config/nextflow.config
+  tags:
+    - dastool
+    - dastool/fastatocontig2bin
+  files:
+    - path: output/dastool/test.tsv
+      md5sum: 6e46c0be14dded7cb13af38f54feea47
+    - path: output/dastool/versions.yml
+      md5sum: ff4b6f14bee4548bf09b5e602c306595
+
+- name: dastool fastatocontig2bin test_dastool_fastatocontig2bin_ungzipped
+  command: nextflow run tests/modules/dastool/fastatocontig2bin -entry test_dastool_fastatocontig2bin_ungzipped -c tests/config/nextflow.config
+  tags:
+    - dastool
+    - dastool/fastatocontig2bin
+  files:
+    - path: output/dastool/test.tsv
+      md5sum: 6e46c0be14dded7cb13af38f54feea47
+    - path: output/dastool/versions.yml