name: "deepbgc_pipeline"
description: detect BGCs in bacterial and fungal genomes using deep learning
keywords:
  - Biosynthetic Gene Cluster
  - deep learning
  - neural network
  - random forest
  - 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']"

input:
  - meta:
      type: map
      description: |
        Groovy Map containing sample information
        e.g. [ id:'test' ]
  - genome:
      type: file
      description: FASTA/GenBank/Pfam CSV file
      pattern: "*.{fasta,fa,fna,gbk,csv}"

output:
  - meta:
      type: map
      description: |
        Groovy Map containing sample information
        e.g. [ id:'test']
  - versions:
      type: file
      description: File containing software versions
      pattern: "versions.yml"
  - readme:
      type: file
      description: txt file containing description of output files
      pattern: "*.{txt}"
  - log:
      type: file
      description: Log output of DeepBGC
      pattern: "*.{txt}"
  - json:
      type: file
      description: AntiSMASH JSON file for sideloading.
      pattern: "*.{json}"
  - bgc_gbk:
      type: file
      description: Sequences and features of all detected BGCs in GenBank format.
      pattern: "*.{bgc.gbk}"
  - bgc_tsv:
      type: file
      description: Table of detected BGCs and their properties.
      pattern: "*.{bgc.tsv}"
  - full_gbk:
      type: file
      description: Fully annotated input sequence with proteins, Pfam domains (PFAM_domain features) and BGCs (cluster features)
      pattern: "*.{full.gbk}"
  - pfam_tsv:
      type: file
      description: Table of Pfam domains (pfam_id) from given sequence (sequence_id) in genomic order, with BGC detection scores.
      pattern: "*.{pfam.tsv}"
  - bgc_png:
      type: file
      description: Detected BGCs plotted by their nucleotide coordinates.
      pattern: "*.{bgc.png}"
  - pr_png:
      type: file
      description: Precision-Recall curve based on predicted per-Pfam BGC scores.
      pattern: "*.{pr.png}"
  - roc_png:
      type: file
      description: ROC curve based on predicted per-Pfam BGC scores.
      pattern: "*.{roc.png}"
  - score_png:
      type: file
      description: BGC detection scores of each Pfam domain in genomic order.
      pattern: "*.{score.png}"

authors:
  - "@louperelo"
  - "@jfy133"