Deeptmhmm (#3981)

* deeptmhmm main.nf

* deeptmhmm meta.yml

* deeptmhmm tests main

* deeptmhmm tests validation

modules/nf-core/deeptmhmm/main.nf

@@ -0,0 +1,62 @@
+process DEEPTMHMM {
+    tag "$meta.id"
+    label 'process_medium'
+
+    conda "bioconda::pybiolib=1.1.1393"
+    container "${ workflow.containerEngine == 'singularity' && !task.ext.singularity_pull_docker_container ?
+        'https://depot.galaxyproject.org/singularity/pybiolib:1.1.1393--pyhdfd78af_0':
+        'biocontainers/pybiolib:1.1.1393--pyhdfd78af_0' }"
+
+    input:
+    tuple val(meta), path(fasta)
+
+    output:
+    tuple val(meta), path("biolib_results/TMRs.gff3")                 , emit: gff3
+    tuple val(meta), path("biolib_results/predicted_topologies.3line"), emit: line3
+    tuple val(meta), path("biolib_results/deeptmhmm_results.md")      , emit: md
+    tuple val(meta), path("biolib_results/*_probs.csv")               , optional: true, emit: csv
+    tuple val(meta), path("biolib_results/plot.png")                  , optional: true, emit: png
+    path "versions.yml"                                               , emit: versions
+
+    when:
+    task.ext.when == null || task.ext.when
+
+    script:
+    def args = task.ext.args ?: ''
+    def is_compressed = fasta.name.endsWith(".gz")
+    def fasta_name = fasta.name.replace(".gz", "")
+
+    """
+    if [ "$is_compressed" == "true" ]; then
+        gzip -c -d $fasta > $fasta_name
+    fi
+
+    biolib \\
+        run \\
+        DTU/DeepTMHMM \\
+        --fasta ${fasta_name} \\
+        $args
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        biolib: \$(echo \$(biolib --version) | sed -n 's/.*version \\([0-9.]*\\).*/\\1/p' )
+    END_VERSIONS
+    """
+
+    stub:
+    def args = task.ext.args ?: ''
+
+    """
+    mkdir biolib_results
+    touch biolib_results/TMRs.gff3
+    touch biolib_results/predicted_topologies.3line
+    touch biolib_results/deeptmhmm_results.md
+    touch biolib_results/MX_probs.csv
+    touch biolib_results/plot.png
+
+    cat <<-END_VERSIONS > versions.yml
+    "${task.process}":
+        biolib: \$(echo \$(biolib --version) | sed -n 's/.*version \\([0-9.]*\\).*/\\1/p' )
+    END_VERSIONS
+    """
+}

modules/nf-core/deeptmhmm/meta.yml

@@ -0,0 +1,67 @@
+---
+# yaml-language-server: $schema=https://raw.githubusercontent.com/nf-core/modules/master/modules/yaml-schema.json
+name: deeptmhmm
+description: A Deep Learning Model for Transmembrane Topology Prediction and Classification
+keywords:
+  - transmembrane
+  - protein
+  - classification
+tools:
+  - deeptmhmm:
+      description: Deep Learning model for Transmembrane Helices protein domain prediction through the BioLib Python Client
+      homepage: https://dtu.biolib.com/DeepTMHMM
+      documentation: https://dtu.biolib.com/DeepTMHMM
+      doi: 10.1101/2022.04.08.487609
+      licence: ["MIT"]
+
+input:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+
+  - fasta:
+      type: file
+      description: Database of sequences in FASTA format
+      pattern: "*.{fasta,fa,fasta.gz,fa.gz}"
+
+output:
+  - meta:
+      type: map
+      description: |
+        Groovy Map containing sample information
+        e.g. [ id:'test', single_end:false ]
+
+  - gff3:
+      type: file
+      description: Predicted topologies (inside, outside, TMhelix) in general Feature Format Version 3
+      pattern: "biolib_results/TMRs.gff3"
+
+  - line3:
+      type: file
+      description: Predicted topologies and information of protein sequences in three lines (name, sequence, topology)
+      pattern: "biolib_results/predicted_topologies.3line"
+
+  - md:
+      type: file
+      description: Markdown results file
+      pattern: "biolib_results/deeptmhmm_results.md"
+
+  - csv:
+      type: file
+      description: CSV file with per-residue predictions for the likelihood of each amino acid being in structural regions such as Beta-sheet, Periplasm, Membrane, Inside, Outside or Signal (only when querying against genomic fasta)
+      pattern: "biolib_results/*_probs.csv"
+
+  - png:
+      type: file
+      description: Most likely topology probability line plots (only when querying against genomic fasta)
+      pattern: "biolib_results/plot.png"
+
+  - versions:
+      type: file
+      description: File containing software versions
+      pattern: "versions.yml"
+
+authors:
+  - "@vagkaratzas"

tests/config/pytest_modules.yml

@@ -962,6 +962,10 @@ deepcell/mesmer:
   - modules/nf-core/deepcell/mesmer/**
   - tests/modules/nf-core/deepcell/mesmer/**
 
+deeptmhmm:
+  - modules/nf-core/deeptmhmm/**
+  - tests/modules/nf-core/deeptmhmm/**
+
 deeptools/bamcoverage:
   - modules/nf-core/deeptools/bamcoverage/**
   - tests/modules/nf-core/deeptools/bamcoverage/**

tests/modules/nf-core/deeptmhmm/main.nf

@@ -0,0 +1,19 @@
+#!/usr/bin/env nextflow
+
+nextflow.enable.dsl = 2
+
+include { DEEPTMHMM } from '../../../../modules/nf-core/deeptmhmm/main.nf'
+
+workflow test_deeptmhmm {
+
+    fasta = [ file(params.test_data['sarscov2']['genome']['proteome_fasta'], checkIfExists: true) ]
+
+    DEEPTMHMM ( [ [id:'test'], fasta ] )
+}
+
+workflow test_deeptmhmm_gz {
+
+    fasta_gz = [ file(params.test_data['sarscov2']['genome']['genome_fasta_gz'], checkIfExists: true) ]
+
+    DEEPTMHMM ( [ [id:'test'], fasta_gz ] )
+}

tests/modules/nf-core/deeptmhmm/nextflow.config

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

tests/modules/nf-core/deeptmhmm/test.yml

@@ -0,0 +1,29 @@
+- name: deeptmhmm test_deeptmhmm
+  command: nextflow run ./tests/modules/nf-core/deeptmhmm -entry test_deeptmhmm -c ./tests/config/nextflow.config
+  tags:
+    - deeptmhmm
+  files:
+    - path: output/deeptmhmm/biolib_results/TMRs.gff3
+      md5sum: 563b7cf9659f955c3353ab7ba2ed4c41
+    - path: output/deeptmhmm/biolib_results/deeptmhmm_results.md
+      md5sum: 14867a44faf50cab3f56874b47ff21ec
+    - path: output/deeptmhmm/biolib_results/predicted_topologies.3line
+      md5sum: 50c853fd0dc0f28aff0bdb6d35d08ca2
+    - path: output/deeptmhmm/versions.yml
+
+- name: deeptmhmm test_deeptmhmm_gz
+  command: nextflow run ./tests/modules/nf-core/deeptmhmm -entry test_deeptmhmm_gz -c ./tests/config/nextflow.config
+  tags:
+    - deeptmhmm
+  files:
+    - path: output/deeptmhmm/biolib_results/MT192765.1_probs.csv
+      md5sum: ac50b0378d09761e9c2931073ee92436
+    - path: output/deeptmhmm/biolib_results/TMRs.gff3
+      md5sum: f338e8a8280d62112ed6d84087dc9f82
+    - path: output/deeptmhmm/biolib_results/deeptmhmm_results.md
+      md5sum: 2873f7df7582bef408ca8e51f5119dd9
+    - path: output/deeptmhmm/biolib_results/plot.png
+      md5sum: 5e9e47bfc5c07b81c1056293d3329920
+    - path: output/deeptmhmm/biolib_results/predicted_topologies.3line
+      md5sum: 95b7fedc0a110dd20ab3e265727c2c62
+    - path: output/deeptmhmm/versions.yml