Deepbgcdownload (#2010)

* deepbgc_download first version

* with test.yml

* Apply suggestions from code review

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

* Apply suggestions from code review

* delete trailing whitespace main.nf

Co-authored-by: James A. Fellows Yates <jfy133@gmail.com>

modules/deepbgc/download/main.nf

@@ -0,0 +1,31 @@
+process DEEPBGC_DOWNLOAD {
+    tag "download"
+    label 'process_low'
+
+    conda (params.enable_conda ? "bioconda::deepbgc=0.1.30" : null)
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/deepbgc:0.1.30--pyhb7b1952_1':
+        'quay.io/biocontainers/deepbgc:0.1.30--pyhb7b1952_1' }"
+
+    output:
+    path "deepbgc_db"             , emit: db
+    path "versions.yml"           , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+
+    """
+    export DEEPBGC_DOWNLOADS_DIR='./deepbgc_db'
+
+    deepbgc \\
+        download
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        deepbgc: \$(echo \$(deepbgc info 2>&1 /dev/null/ | grep 'version' | cut -d " " -f3) )
+    END_VERSIONS
+    """
+}

modules/deepbgc/download/meta.yml

@@ -0,0 +1,34 @@
+name: "deepbgc_download"
+description: database to detect BGCs in bacterial and fungal genomes using deep learning
+keywords:
+  - database
+  - download
+  - BGC
+  - Biosynthetic Gene Cluster
+  - deep learning
+  - neural network
+  - random forrest
+  - genomes
+  - bacteria
+  - fungi
+tools:
+  - "deepbgc":
+      description: "DeepBGC - Biosynthetic Gene Cluster detection and classification"
+      homepage: "https://github.com/Merck/deepbgc"
+      documentation: "https://github.com/Merck/deepbgc"
+      tool_dev_url: "https://github.com/Merck/deepbgc"
+      doi: "10.1093/nar/gkz654"
+      licence: "['MIT']"
+
+output:
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+  - deepbgc_db:
+      type: directory
+      description: contains files of reference db from 'deepbgc download'
+      pattern: "*db"
+
+authors:
+  - "@louperelo"

tests/config/pytest_modules.yml

@@ -599,6 +599,10 @@ deeparg/predict:
   - modules/deeparg/predict/**
   - tests/modules/deeparg/predict/**
 
+deepbgc/download:
+  - modules/deepbgc/download/**
+  - tests/modules/deepbgc/download/**
+
 deeptools/bamcoverage:
   - modules/deeptools/bamcoverage/**
   - tests/modules/deeptools/bamcoverage/**

tests/modules/deepbgc/download/main.nf

@@ -0,0 +1,10 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { DEEPBGC_DOWNLOAD } from '../../../../modules/deepbgc/download/main.nf'
+
+workflow test_deepbgc_download {
+
+    DEEPBGC_DOWNLOAD (  )
+}

tests/modules/deepbgc/download/nextflow.config

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

tests/modules/deepbgc/download/test.yml

@@ -0,0 +1,30 @@
+- name: deepbgc download test_deepbgc_download
+  command: nextflow run ./tests/modules/deepbgc/download -entry test_deepbgc_download -c ./tests/config/nextflow.config  -c ./tests/modules/deepbgc/download/nextflow.config
+  tags:
+    - deepbgc
+    - deepbgc/download
+  files:
+    - path: output/deepbgc/deepbgc_db/0.1.0/classifier/product_activity.pkl
+      md5sum: 90f0c010460e9df882cb057664a49f30
+    - path: output/deepbgc/deepbgc_db/0.1.0/classifier/product_class.pkl
+      md5sum: f78a2eda240403d2f40643d42202f3ac
+    - path: output/deepbgc/deepbgc_db/0.1.0/detector/clusterfinder_geneborder.pkl
+      md5sum: ca4be7031ae9f70780f17c616a4fa5b5
+    - path: output/deepbgc/deepbgc_db/0.1.0/detector/clusterfinder_original.pkl
+      md5sum: 2ca2429bb9bc99a401d1093c376b37aa
+    - path: output/deepbgc/deepbgc_db/0.1.0/detector/clusterfinder_retrained.pkl
+      md5sum: 65679a3b61c562ff4b84bdb574bb6d93
+    - path: output/deepbgc/deepbgc_db/0.1.0/detector/deepbgc.pkl
+      md5sum: 7e9218be79ba45bc9adb23bed3845dc1
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.clans.tsv
+      md5sum: a0a4590ffb2b33b83ef2b28f6ead886b
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.hmm
+      md5sum: 79a3328e4c95b13949a4489b19959fc5
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.hmm.h3f
+      md5sum: cbca323cf8dd4e5e7c109114ec444162
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.hmm.h3i
+      md5sum: 5242332a3f6a60cd1ab634cd9331afd6
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.hmm.h3m
+      md5sum: 1fe946fa2b3bcde1d4b2bad732bce612
+    - path: output/deepbgc/deepbgc_db/common/Pfam-A.31.0.hmm.h3p
+      md5sum: 27b98a1ded123b6a1ef72db01927017c