comparison data_manager_build_alfa_indexes/data_manager/data_manager_build_alfa_indexes.py @ 24:424d6204a298 draft

Uploaded

23:c7dbef0e9ed7

import urllib2
import subprocess
import gzip
import os
import tempfile
import logging
from optparse import OptionParser
from galaxy.util.json import from_json_string, to_json_string

def get_arg():
    parser = OptionParser()

    grep_result = subprocess.Popen(['grep', species_name, list_species_file_name], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    species_lines_matched, grep_error = grep_result.communicate()
    if grep_error != None or species_lines_matched == "":
        msg = 'The species \'%s\' is not referenced on Ensembl (%s)' % (species_name, kingdom)
        logging.critical(msg)
        sys.exit(msg)

    species_lines = species_lines_matched.split('\n')
    del species_lines[-1]
    nb_lines = len(species_lines)

            exact_match = re.search('^%s$' % species_name, list_species[i])
            if exact_match:
                print("-> Referenced !\n")
                return species_name, species_lines[i]
        msg = 'The string \'%s\' has been matched against the list of Ensembl Species but is not a complete species name.\nPlease retry with one of the following species names:\n%s' % (species_name, list_species[0:])
        logging.critical(msg)
        sys.exit(msg)

def get_ensembl_collection(kingdom, species_line):
    print("*** Extracting the %s_collection of the species" % kingdom)
    collection_regex = re.compile('%s_.+_collection' % kingdom.lower())

    print("*** Generating alfa indexes from %s" % gtf_file_name)
    alfa_result = subprocess.Popen(['python', path_to_alfa, '-a',  gtf_file_name], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    alfa_out, alfa_err = alfa_result.communicate()
    if alfa_err != None and not re.search('### End of program', alfa_err):
        msg = 'Generation Failed due an alfa error: %s' % (alfa_err)
        logging.critical(msg)
        sys.exit(msg)
    print("Alfa prompt:\n%s" % alfa_out)
    print("-> Generated !\n")

def get_data_table_new_entry(gtf_archive_name):

    path_to_alfa = os.path.join(tool_dir, 'ALFA.py')

    if options.output_filename == None:
        msg = 'No json output file specified'
        logging.critical(msg)
        sys.exit(msg)
    output_filename = options.output_filename
    params = from_json_string(open(output_filename).read())
    target_directory = params['output_data'][0]['extra_files_path']
    os.mkdir(target_directory)

24:424d6204a298

import urllib2
import subprocess
import gzip
import os
import tempfile
from optparse import OptionParser
from galaxy.util.json import from_json_string, to_json_string

def get_arg():
    parser = OptionParser()

    grep_result = subprocess.Popen(['grep', species_name, list_species_file_name], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    species_lines_matched, grep_error = grep_result.communicate()
    if grep_error != None or species_lines_matched == "":
        msg = 'The species \'%s\' is not referenced on Ensembl (%s)' % (species_name, kingdom)
        sys.exit(msg)

    species_lines = species_lines_matched.split('\n')
    del species_lines[-1]
    nb_lines = len(species_lines)

            exact_match = re.search('^%s$' % species_name, list_species[i])
            if exact_match:
                print("-> Referenced !\n")
                return species_name, species_lines[i]
        msg = 'The string \'%s\' has been matched against the list of Ensembl Species but is not a complete species name.\nPlease retry with one of the following species names:\n%s' % (species_name, list_species[0:])
        sys.exit(msg)

def get_ensembl_collection(kingdom, species_line):
    print("*** Extracting the %s_collection of the species" % kingdom)
    collection_regex = re.compile('%s_.+_collection' % kingdom.lower())

    print("*** Generating alfa indexes from %s" % gtf_file_name)
    alfa_result = subprocess.Popen(['python', path_to_alfa, '-a',  gtf_file_name], stdout=subprocess.PIPE, stderr=subprocess.STDOUT)
    alfa_out, alfa_err = alfa_result.communicate()
    if alfa_err != None and not re.search('### End of program', alfa_err):
        msg = 'Generation Failed due an alfa error: %s' % (alfa_err)
        sys.exit(msg)
    print("Alfa prompt:\n%s" % alfa_out)
    print("-> Generated !\n")

def get_data_table_new_entry(gtf_archive_name):

    path_to_alfa = os.path.join(tool_dir, 'ALFA.py')

    if options.output_filename == None:
        msg = 'No json output file specified'
        sys.exit(msg)
    output_filename = options.output_filename
    params = from_json_string(open(output_filename).read())
    target_directory = params['output_data'][0]['extra_files_path']
    os.mkdir(target_directory)