Add MOTUS/MERGE (#1734)

* Add motus/merge module, missing test yml

* Fix prefix

* Fix polymersase

* Update test meta.yml

* Try new version system

* Typo fix

* Clairfy docs for the version file

* Switch to directory auto detection

* Now with the change and not just meta

* Remove now unneeded input parameter from test

* Fix input type check

* Try adding db requirement to fix docker issue

* Update modules/motus/merge/main.nf

* Fix input

* Apply suggestions from code review

* Fix md5sums

* Replace debugging db location with actual

* Fix contains due to further varialbility

* Indenting

modules/motus/downloaddb/main.nf

@@ -33,7 +33,7 @@ process MOTUS_DOWNLOADDB {
     ## mOTUs will check the database version is same version as exec version.
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        mOTUs: \$(grep motus db_mOTU/db_mOTU_versions | sed 's/motus\\t//g')
+        motus: \$(grep motus db_mOTU/db_mOTU_versions | sed 's/motus\\t//g')
     END_VERSIONS
     """
 }

modules/motus/merge/main.nf

@@ -0,0 +1,47 @@
+VERSION = '3.0.1'
+
+process MOTUS_MERGE {
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::motus=3.0.1" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/motus:3.0.1--pyhdfd78af_0':
+        'quay.io/biocontainers/motus:3.0.1--pyhdfd78af_0' }"
+
+    input:
+    path input
+    path db // to stop docker saying it can't find it... would have to have the module in upstream steps anyway
+    path profile_version_yml, stageAs: 'profile_version.yml'
+    val biom_format
+
+    output:
+    path("*.txt") , optional: true, emit: txt
+    path("*.biom"), optional: true, emit: biom
+    path "versions.yml" , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = 'motus_merged'
+    def cmd_input = input.size() > 1 ? "-i ${input.join(',')}" : input.isDirectory() ? "-d ${input}" : "-i ${input}"
+    def output = biom_format ? "-B -o ${prefix}.biom" : "-o ${prefix}.txt"
+    """
+    motus \\
+        merge \\
+        -db $db \\
+        ${cmd_input} \\
+        $args \\
+        ${output}
+
+    ## Take version from the mOTUs/profile module output, as cannot reconstruct
+    ## version without having database staged in this directory.
+    VERSION=\$(cat ${profile_version_yml} | grep '/*motus:.*' | sed 's/.*otus: //g')
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        motus: \$VERSION
+    END_VERSIONS
+    """
+}

modules/motus/merge/meta.yml

@@ -0,0 +1,57 @@
+name: "motus_merge"
+description: Taxonomic meta-omics profiling using universal marker genes
+keywords:
+  - classify
+  - metagenomics
+  - fastq
+  - taxonomic profiling
+  - merging
+  - merge
+  - otu table
+tools:
+  - "motus":
+      description: "Marker gene-based OTU (mOTU) profiling"
+      homepage: "https://motu-tool.org/"
+      documentation: "https://github.com/motu-tool/mOTUs/wiki"
+      tool_dev_url: "https://github.com/motu-tool/mOTUs"
+      doi: "10.1038/s41467-019-08844-4"
+      licence: "['GPL v3']"
+
+input:
+  - input:
+      type: file
+      description: |
+        List of output files (more than one) from motus profile,
+        or a single directory containing motus output files.
+  - db:
+      type: directory
+      description: |
+        mOTUs database downloaded by `motus downloadDB`
+        pattern: "db_mOTU/"
+  - profile_version_yml:
+      type: file
+      description: |
+        A single versions.yml file output from motus/profile. motus/merge cannot reconstruct
+        this itself without having the motus database present and configured with the tool
+        so here we take it from what is already reported by the upstream module.
+      pattern: "versions.yml"
+  - biom_format:
+      type: boolean
+      description: Whether to save output OTU table in biom format
+
+output:
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - txt:
+      type: file
+      description: OTU table in txt format, if BIOM format not requested
+      pattern: "*.txt"
+  - biom:
+      type: file
+      description: OTU table in biom format, if BIOM format requested
+      pattern: "*.biom"
+
+authors:
+  - "@jfy133"

modules/motus/profile/main.nf

@@ -48,7 +48,7 @@ process MOTUS_PROFILE {
     fi
     cat <<-END_VERSIONS > versions.yml
     "${task.process}":
-        mOTUs: \$VERSION
+        motus: \$VERSION
     END_VERSIONS
     """
 }

tests/config/pytest_modules.yml

@@ -1470,6 +1470,10 @@ motus/downloaddb:
   - modules/motus/downloaddb/**
   - tests/modules/motus/downloaddb/**
 
+motus/merge:
+  - modules/motus/merge/**
+  - tests/modules/motus/merge/**
+
 motus/profile:
   - modules/motus/profile/**
   - tests/modules/motus/profile/**

tests/modules/motus/merge/main.nf

@@ -0,0 +1,29 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { MOTUS_DOWNLOADDB } from '../../../../modules/motus/downloaddb/main.nf'
+include { MOTUS_PROFILE    } from '../../../../modules/motus/profile/main.nf'
+include { MOTUS_MERGE } from '../../../../modules/motus/merge/main.nf'
+
+workflow test_motus_merge {
+
+    input = Channel.fromList(
+        [
+            [
+                [ id:'test', single_end:true ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+            ],
+            [
+                [ id:'test2', single_end:true ], // meta map
+                file(params.test_data['sarscov2']['illumina']['test_1_fastq_gz'], checkIfExists: true)
+            ]
+        ]
+    )
+
+    MOTUS_DOWNLOADDB ( file('https://raw.githubusercontent.com/motu-tool/mOTUs/master/motus/downloadDB.py') )
+    MOTUS_PROFILE ( input, MOTUS_DOWNLOADDB.out.db )
+    MOTUS_MERGE ( MOTUS_PROFILE.out.out.map{it[1]}.collect(), MOTUS_DOWNLOADDB.out.db, MOTUS_PROFILE.out.versions.first(), false )
+
+}
+

tests/modules/motus/merge/nextflow.config

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

tests/modules/motus/merge/test.yml

@@ -0,0 +1,12 @@
+- name: motus merge test_motus_merge
+  command: nextflow run ./tests/modules/motus/merge -entry test_motus_merge -c ./tests/config/nextflow.config  -c ./tests/modules/motus/merge/nextflow.config
+  tags:
+    - motus
+    - motus/merge
+  files:
+    - path: output/motus/motus_merged.txt
+      contains: ["#consensus_taxonomy"]
+    - path: output/motus/test.out
+      contains: ["#consensus_taxonomy"]
+    - path: output/motus/test2.out
+      contains: ["#consensus_taxonomy"]