def generate_referenced_species_database(protein_id, referenced_species):
'''
Creates a database for a referenced species and protein, using formatdb
@param protein_id
@param referenced_species
'''
logger = Logger.Instance()
alignment_logger = logger.get_logger('alignment')
command_generator = CommandGenerator()
crawler = DirectoryCrawler()
exon_container = ExonContainer.Instance()
input_exons = exon_container.get((protein_id, referenced_species, "ensembl"))
#source_exon_file = "{0}/{1}.fa".format(crawler.get_exon_ensembl_path(protein_id), referenced_species)
input_db_file = "{0}/{1}.fa".format(crawler.get_database_path(protein_id), referenced_species)
sequence_type = "Nucleotide"
input_exons.export_coding_exons_to_fasta(input_db_file)
command = command_generator.generate_formatdb_command(input_db_file, sequence_type)
command_return = Popen(command, shell=True, stdin=PIPE, stdout=PIPE, stderr=STDOUT, close_fds=True)
output = command_return.stdout.read()
if output != "":
#LOGGING
alignment_logger.warning("{0}, {1}, REF SPECIES DB, {2}".format(protein_id, referenced_species.strip(), output.strip()))
return False
return True
def generate_blastn_alignments(protein_id, species_list = None, referenced_species = "Homo_sapiens"):
'''
Runs the blastn program for a specified protein and list of species
@param protein_id
@param species_list: if provided, runs blastn for this list of species, \
otherwise runs for species that are missing the blastn output \
who are determined by .status file in the blastn folder.
'''
logger = Logger.Instance()
alignment_logger = logger.get_logger('alignment')
crawler = DirectoryCrawler()
command_generator = CommandGenerator()
alignment_generator = AlignmentTargetGenerator()
failed_species_list = []
# retrieve the blastn targets
if (not species_list):
species_list = alignment_generator.get_blastn_targets(protein_id)
for species in species_list:
############# MOVE TO ANOTHER FNC
output_file = "{0}/{1}.blastout".format(crawler.get_blastn_path(protein_id), species.strip())
input_file = "{0}/{1}.fa".format(crawler.get_expanded_gene_path(protein_id), species.strip())
database = "{0}/{1}.fa".format(crawler.get_database_path(protein_id), referenced_species)
command = command_generator.generate_blastn_command(database, input_file, output_file)
command_return = Popen(command, shell=True, stdin=PIPE, stdout=PIPE, stderr=STDOUT, close_fds=True)
output = command_return.stdout.read()
if output != "":
#LOGGING
os.remove(output_file)
alignment_logger.warning("{0}, {1}, BLASTN, {2}".format(protein_id, species.strip(), output.strip()))
failed_species_list.append(species.strip())
if failed_species_list:
alignment_generator.set_failed_blastn_targets(protein_id, failed_species_list)
return False
return True
def generate_SW_gene_alignments(protein_id, species_list = None, referenced_species = "Homo_sapiens"):
'''
Runs the SW program for a specified protein and list of species, using the expanded gene region.
@param protein_id
@param species_list: if provided, runs SW for this list of species, \
otherwise runs for species that are missing the SW output \
who are determined by .status file in the /SW/gene folder.
'''
logger = Logger.Instance()
alignment_logger = logger.get_logger('alignment')
alignment_generator = AlignmentTargetGenerator()
crawler = DirectoryCrawler()
command_generator = CommandGenerator()
if (not species_list):
species_list = alignment_generator.get_SW_gene_targets(protein_id)
failed_species_list = []
for species in species_list:
########### MOVE
output_file = "{0}/{1}.swout".format(crawler.get_SW_gene_path(protein_id), species.strip())
query_sequence_file = "{0}/{1}.fa".format(crawler.get_expanded_gene_path(protein_id), species.strip())
target_fasta_db_file = "{0}/{1}.fa".format(crawler.get_database_path(protein_id), referenced_species)
command = command_generator.generate_SW_command(query_sequence_file, target_fasta_db_file, output_file)
command_return = Popen(command, shell=True, stdin=PIPE, stdout=PIPE, stderr=STDOUT, close_fds=True)
output = command_return.stdout.read()
if output != "":
#LOGGING
alignment_logger.warning("{0}, {1}, SW GENE, {2}".format(protein_id, species.strip(), output.strip()))
failed_species_list.append(species.strip())
os.remove(".sw_stdout_supressed")
if failed_species_list:
alignment_generator.set_failed_SW_gene_targets(protein_id, failed_species_list)
return False
return True