From 53b324281f9bdf0b8aa7aeae8410a65a7c5a0553 Mon Sep 17 00:00:00 2001
From: Felipe Marques de Almeida <felipemarques89@gmail.com>
Date: Tue, 1 Feb 2022 11:47:01 +0100
Subject: [PATCH] Raven assembler (#1087)

Adding a new module for [Raven](https://github.com/lbcb-sci/raven), a De novo genome assembler for long uncorrected reads.
---
 modules/raven/main.nf               | 39 ++++++++++++++++++++++++
 modules/raven/meta.yml              | 47 +++++++++++++++++++++++++++++
 tests/config/pytest_modules.yml     |  4 +++
 tests/modules/raven/main.nf         | 15 +++++++++
 tests/modules/raven/nextflow.config |  5 +++
 tests/modules/raven/test.yml        |  9 ++++++
 6 files changed, 119 insertions(+)
 create mode 100644 modules/raven/main.nf
 create mode 100644 modules/raven/meta.yml
 create mode 100644 tests/modules/raven/main.nf
 create mode 100644 tests/modules/raven/nextflow.config
 create mode 100644 tests/modules/raven/test.yml

diff --git a/modules/raven/main.nf b/modules/raven/main.nf
new file mode 100644
index 00000000..0d81b94b
--- /dev/null
+++ b/modules/raven/main.nf
@@ -0,0 +1,39 @@
+process RAVEN {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda (params.enable_conda ? "bioconda::raven-assembler=1.6.1" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/raven-assembler:1.6.1--h2e03b76_0' :
+        'quay.io/biocontainers/raven-assembler:1.6.1--h2e03b76_0' }"
+
+    input:
+    tuple val(meta), path(reads)
+
+    output:
+    tuple val(meta), path("*.fasta.gz"), emit: fasta
+    tuple val(meta), path("*.gfa.gz")  , emit: gfa
+    path "versions.yml"                , emit: versions
+
+    script:
+    def args = task.ext.args ?: ''
+    def prefix = task.ext.prefix ?: "${meta.id}"
+    """
+    # run tool
+    raven \\
+        -t $task.cpus \\
+        --graphical-fragment-assembly ${prefix}.gfa \\
+        $args \\
+        $reads | \\
+        gzip -c > ${prefix}.fasta.gz
+
+    # compress assembly graph
+    gzip -c ${prefix}.gfa > ${prefix}.gfa.gz
+
+    # get tool version
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        raven: \$( raven --version )
+    END_VERSIONS
+    """
+}
diff --git a/modules/raven/meta.yml b/modules/raven/meta.yml
new file mode 100644
index 00000000..644907a9
--- /dev/null
+++ b/modules/raven/meta.yml
@@ -0,0 +1,47 @@
+name: raven
+description: De novo genome assembler for long uncorrected reads.
+keywords:
+  - de novo
+  - assembly
+  - genome
+  - genome assembler
+  - long uncorrected reads
+tools:
+  - raven:
+      description: Raven is a de novo genome assembler for long uncorrected reads.
+      homepage: https://github.com/lbcb-sci/raven
+      documentation: https://github.com/lbcb-sci/raven#usage
+      tool_dev_url: https://github.com/lbcb-sci/raven
+      doi: doi.org/10.1038/s43588-021-00073-4
+      licence: ['MIT']
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', suffix:'bacteria' ]
+  - reads:
+      type: file
+      description: Input file in FASTA/FASTQ format.
+      pattern: "*.{fasta,fastq,fasta.gz,fastq.gz,fa,fq,fa.gz,fq.gz}"
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', suffix:'bacteria' ]
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - fasta:
+      type: file
+      description: Assembled FASTA file
+      pattern: "*.fasta.gz"
+  - gfa:
+      type: file
+      description: Repeat graph
+      pattern: "*.gfa.gz"
+
+authors:
+  - "@fmalmeida"
diff --git a/tests/config/pytest_modules.yml b/tests/config/pytest_modules.yml
index 06ef1649..fe39bcc5 100644
--- a/tests/config/pytest_modules.yml
+++ b/tests/config/pytest_modules.yml
@@ -1205,6 +1205,10 @@ rasusa:
   - modules/rasusa/**
   - tests/modules/rasusa/**
 
+raven:
+  - modules/raven/**
+  - tests/modules/raven/**
+
 raxmlng:
   - modules/raxmlng/**
   - tests/modules/raxmlng/**
diff --git a/tests/modules/raven/main.nf b/tests/modules/raven/main.nf
new file mode 100644
index 00000000..01b04519
--- /dev/null
+++ b/tests/modules/raven/main.nf
@@ -0,0 +1,15 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { RAVEN } from '../../../modules/raven/main.nf'
+
+workflow test_raven {
+    
+    input = [ 
+              [ id:'test', single_end:false ], // meta map
+              [ file(params.test_data['bacteroides_fragilis']['nanopore']['test_fastq_gz'], checkIfExists: true) ] 
+            ]
+
+    RAVEN ( input )
+}
diff --git a/tests/modules/raven/nextflow.config b/tests/modules/raven/nextflow.config
new file mode 100644
index 00000000..8730f1c4
--- /dev/null
+++ b/tests/modules/raven/nextflow.config
@@ -0,0 +1,5 @@
+process {
+
+    publishDir = { "${params.outdir}/${task.process.tokenize(':')[-1].tokenize('_')[0].toLowerCase()}" }
+
+}
diff --git a/tests/modules/raven/test.yml b/tests/modules/raven/test.yml
new file mode 100644
index 00000000..bbede642
--- /dev/null
+++ b/tests/modules/raven/test.yml
@@ -0,0 +1,9 @@
+- name: raven test_raven
+  command: nextflow run ./tests/modules/raven -entry test_raven -c ./tests/config/nextflow.config -c ./tests/modules/raven/nextflow.config
+  tags:
+    - raven
+  files:
+    - path: output/raven/test.fasta.gz
+      md5sum: b026b6aba793a9f2bbfb17f732c91926
+    - path: output/raven/test.gfa.gz
+      md5sum: 09053490495f6deb7a3941e559cda626