speciesData.py

#!/usr/bin/env python3
# -*- coding: utf-8 -*-

import os
import sys
import utilities

from _datetime import datetime


class SpeciesData:
    """
    This class contains attributes and functions to interact with the galaxy container of the GGA environment
    Parent class of LoadData, DeploySpeciesStack and RunWorkflow

    """

    def __init__(self, parameters_dictionary):
        self.config_dictionary = None
        self.parameters_dictionary = parameters_dictionary
        self.species = parameters_dictionary["description"]["species"]
        self.genus = parameters_dictionary["description"]["genus"]
        self.strain = parameters_dictionary["description"]["strain"]
        self.sex = parameters_dictionary["description"]["sex"]
        self.common = parameters_dictionary["description"]["common_name"]
        self.date = datetime.today().strftime("%Y-%m-%d")
        self.origin = parameters_dictionary["description"]["origin"]
        self.performed = parameters_dictionary["data"]["performed_by"]
        if parameters_dictionary["data"]["genome_version"] == "":
            self.genome_version = "1.0"
        else:
            self.genome_version = parameters_dictionary["data"]["genome_version"]
        if parameters_dictionary["data"]["ogs_version"] == "":
            self.ogs_version = "1.0"
        else:
            self.ogs_version = parameters_dictionary["data"]["ogs_version"]
        self.genus_lowercase = self.genus[0].lower() + self.genus[1:]
        self.genus_uppercase = self.genus[0].upper() + self.genus[1:]
        self.chado_species_name = "{0} {1}".format(self.species, self.sex)
        self.species_folder_name = "_".join([self.genus_lowercase, self.species, self.strain, self.sex])
        self.full_name = ' '.join(utilities.filter_empty_not_empty_items([self.genus_uppercase, self.species, self.strain, self.sex])["not_empty"])
        self.full_name_lowercase = self.full_name.lower()
        self.abbreviation = "_".join(utilities.filter_empty_not_empty_items([self.genus_lowercase[0], self.species, self.strain, self.sex])["not_empty"])
        self.genus_species = self.genus_lowercase + "_" + self.species
        self.instance_url = None
        self.instance = None
        self.history_id = None
        self.library = None
        self.library_id = None
        self.script_dir = os.path.dirname(os.path.realpath(sys.argv[0]))
        self.main_dir = None
        self.species_dir = None
        self.org_id = None
        self.genome_analysis_id = None
        self.ogs_analysis_id = None
        self.tool_panel = None
        self.datasets = dict()
        self.source_files = dict()
        self.workflow_name = None
        self.metadata = dict()
        self.api_key = None
        self.datasets = dict()
        # API key used to communicate with the galaxy instance. Cannot be used to do user-tied actions
        self.config = None
        # Custom config used to set environment variables inside containers, defaults to the one in the repo
        if parameters_dictionary["data"]["parent_directory"] == "" or parameters_dictionary["data"]["parent_directory"] == "/path/to/closest/parent/dir": self.source_data_dir = "/shared/projects/phaeoexplorer/"  # Testing path for phaeoexplorer data
        else:
            self.source_data_dir = parameters_dictionary["data"]["parent_directory"]
        self.species_folder_name = "_".join(utilities.filter_empty_not_empty_items([self.genus_lowercase, self.species, self.strain, self.sex])["not_empty"])
        self.existing_folders_cache = {}
        self.bam_metadata_cache = {}
        # # Directory/subdirectories where data files are located (fasta, gff, ...)
        # self.do_update = False
        # Update the instance (in histories corresponding to the input) instead of creating a new one
        # self.species_name_regex_litteral = "(?=\w*V)(?=\w*A)(?=\w*R)(?=\w*I)(?=\w*A)(?=\w*B)(?=\w*L)(?=\w*E)\w+"
        #  Placeholder regex for file matching