RegGTFExtractor.py

#!/usr/bin/env python3

import argparse
from Modules.Ensembl.Ensembl import Ensembl
from Modules.ucsc.ucsc import UcscGtf
from Modules.Uniquifier import UniqueFilter
from Modules.SaveResults import ResultSaver
import os
import json


def check_for_local_folder(wd):

    if not os.path.isdir(os.path.join(wd+"/EnsemblData")):

        os.mkdir(os.path.join(wd+"/EnsemblData"))

    if not os.path.isdir(os.path.join(wd+"/UCSCData" )):
        os.mkdir(os.path.join(wd+"/UCSCData" ))


def check_filter(tissue_cmd, org, wd):
    path_to_config = os.path.join(wd + "/config/celltypes_" + org + ".json" )
    tissues_config = []
    if not tissue_cmd:
        return False
    with open(path_to_config) as input_file:
        data = json.loads(input_file.read())
        for x in data:
            tissues_config.append(x["type"])

    if any(tissue in tissues_config for tissue in tissue_cmd):
        return True

    else:
        return False


def main_script(org, wd, tissuetype=None):

    check_for_local_folder(wd)
    if check_filter(tissuetype, org, wd):
        tissues = tissuetype
        print("Filter detected !")
    else:
        tissues = None
        print("Filter not detected !")

    ucsc = UcscGtf(org, wd)
    ense = Ensembl(org, wd)
    print("Getting Unique Results")
    unique_filter = UniqueFilter(ense.get_gtf(), ucsc.get_gtf(), tissues)
    ResultSaver(unique_filter.get_results(), org, tissues)


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='GTF-Generator from UCSC Table Browser and Ensembl Regulatory Build' )
    parser.add_argument('organism', help='Source organism [ homo_sapiens or mus_musculus ]', action='store', nargs='?', type=str)
    parser.add_argument('--tissue', help='Tissue- or Celltype(s)', action='store', nargs='*', type=str)
    parser.add_argument('--wd', help='Working directory. default: "."', action='store', default='.', type=str)
    args = vars(parser.parse_args())
    print("Working Dir: " + args["wd"])
    if args["organism"]:
        print("Working Dir: " + args["wd"])
        main_script(args["organism"], args["wd"], args["tissue"])
    else:
        print("No Arguments found -> See python3 ./RegGTFExtractor.py -h for help.")