ActivityTable.py

import os.path
from Modules.Ensembl.ActivityTableGenerator import ATGenerator


class ActivityTable:

    """
    Class for checking activity_table and generating them.
    activityTable = byte Representation of activity status
    corresponding to the generator schema default:

    0, "activity=ACTIVE",
    1, "activity=POISED",
    2, "activity=REPRESSED",
    3, "activity=INACTIVE",
    4, "activity=NA"

    @author: Sebastian Beyvers
    @contact: sebastian.beyvers@med.uni-giessen.de

    """

    def __init__(self, organism, current_release, wd, data_dir):

        # Constructor for the ActivityTable-Class
        # input_parameter: organism        = input organism
        #                  current_release = current used Ensembl release
        #                  wd              = working dir (default working directory, data_dir is used if specified)
        #                  data_dir        = data directory (this is used as directory if specified)

        if data_dir:
            self.link = os.path.join(data_dir + "/EnsemblData/", current_release, organism, "activity")
        else:
            self.link = os.path.join(wd + "/data/EnsemblData/", current_release, organism, "activity")
        self.folders = next(os.walk(self.link))[1]

        # List to represent Index with activitystatus for ATGenerator class

        self.generator = ATGenerator(["activity=ACTIVE",
                               "activity=POISED",
                               "activity=REPRESSED",
                               "activity=INACTIVE",
                               "activity=NA"])

    def check_and_generate_activity_table(self):

        # checks if file (table.bin) already exists for celltype -> generates new one if missing

        for subfolder in self.folders:
            folder_link = os.path.join(self.link, subfolder)
            sf_link = os.path.join(folder_link, "table.bin")

            # If table.bin is missing:

            if not os.path.isfile(sf_link):
                print("No ActivityTable for "+subfolder+" found, generating new one.")
                self.generate_table(folder_link)

            # Else: Do nothing

        print("All ActivityTables found, proceeding")

    def generate_table(self, link):

        # generates the table and saves it as table.bin file
        # input_parameter: link = link to ensembl activity folder for specific celltype
        # generates table.bin file in link folder

        for root, dirs, files in os.walk(link):
            for file in files:
                if file.endswith(".gff.gz"):
                    originpath = os.path.join(root, file)
                    file_path = os.path.join(root, "table.bin")
                    with open(file_path, "wb") as f:
                        f.write(self.generator.read_table(originpath))
                    print("New ActivityTable generated in: " + root)


# Debug
# e = ActivityTable("homo_sapiens", "release-94")
# e.check_and_generate_activity_table()