input_example.yml

# Input file for the automated creation GGA docker stacks
# The file consists in a "list" of species for which the script will have to create these stacks/load data into galaxy/run workflows
# This file is internally turned into a list of dictionaries by the scripts

ectocarpus_sp2_male:  # Dummy value the user gives to designate the species (isn't used by the script)
  description:
  # Species description, leave blank if unknown or you don't want it to be used
  # These parameters are used to set up the various urls and adresses in different containers
  # The script requires at least the genus to be specified
    genus: "ectocarpus"  # Mandatory!
    species: "sp2" # # Mandatory!
    sex: "male"
    strain: ""
    common_name: ""
    origin: ""
  # the sex and strain, the script will look for files containing the genus, species, sex and strain of the species)
  # If no file corresponding to the description is found, this path will be considered empty and the script will
  # proceed to the next step (create the directory tree for the GGA docker stack)
  data:
  # Sequence of paths to the different datasets to copy and import into galaxy
  # Paths must be absolute paths
    genome_path: "/path/to/fasta" # Mandatory!
    transcripts_path: "/path/to/fasta"  # Mandatory!
    proteins_path: "/path/to/fasta" # Mandatory!
    gff_path: "/path/to/gff" # Mandatory!
    interpro_path: "/path/to/interpro"
    orthofinder_path: "/path/to/orthofinder"
    blastp_path: "/path/to/blastx"
    blastx_path: "/path/to/blastp"
    # If the user has several datasets of the same 'nature' (gff, genomes, ...) to upload to galaxy, the next scalar is used by the script to differentiate
    # between these different versions and name directories according to it and not overwrite the existing data
    # If left empty, the genome will be considered version "1.0"
    genome_version: "1.0"
    # Same as genome version, but for the OGS analysis
    ogs_version: ""
    performed_by: ""
  services:
  # Describe what optional services to deploy for the stack
  # By default, only tripal, tripaldb and galaxy services will be deployed
    blast: "False"
    wiki: "False"
    apollo: "False"

# Second example without the explanation
ectocarpus_sp2_female:
  description:
    genus: "ectocarpus"
    species: "sp2"
    sex: "female"
    strain: ""
    common_name: ""
    origin: ""
  data:
    genome_path: "/path/to/fasta"
    transcripts_path: "/path/to/fasta"
    proteins_path: "/path/to/fasta"
    gff_path: "/path/to/gff"
    interpro_path: "/path/to/interpro"
    orthofinder_path: "/path/to/orthofinder"
    blastp_path: "/path/to/blastx"
    blastx_path: "/path/to/blastp"
    genome_version: "1.0"
    ogs_version: "1.0"
    performed_by: ""
  services:
    blast: "False"
    wiki: "False"
    apollo: "False"