def main():
"""
The program - The work horse
:return: success
"""
path = os.path.abspath("data/genomes/test/")
output_directory = os.path.abspath("data/logger")
print(path)
# setup the application logging
LOG = logging_functions.create_logger()
fh = logging.FileHandler(
os.path.join(output_directory, 'pans_labyrinth.log'), 'w', 'utf-8')
LOG.addHandler(fh)
#options = commandline.arg_parser(client)
#dgraph.execute_args(client, options)
#LOG.debug(options)
LOG.info("Starting pans_labyrinth")
stub = dgraph.create_client_stub()
client = dgraph.create_client(stub)
dgraph.drop_all(client)
dgraph.add_schema(client)
LOG.info("Starting to create graph")
for filepath in files.walkdir(path):
with open(filepath, 'rb') as file:
dgraph.create_graph(client, file, filepath)
stub.close()
LOG.info("ALL DONE")
def test_non_fasta():
"""
Function to test that the program only accepts .fasta files. Could
create issues if the program was reading non-fasta files and inserting the
values into the graph.
"""
path = os.path.abspath("pans_labyrinth/tests/data/bad_fasta")
with pytest.raises(SystemExit) as se:
stub = dgraph.create_client_stub()
client = dgraph.create_client(stub)
dgraph.drop_all(client)
dgraph.add_schema(client)
for filepath in files.walkdir(path):
with open(filepath, 'rb') as file:
dgraph.create_graph(client, file, filepath)
assert se.type == SystemExit
def test_no_file():
"""
Function to ensure that if no file is present in the given directory then
The program fails.
"""
path = os.path.abspath("pans_labyrinth/tests/data/empty_test")
with pytest.raises(SystemExit) as se:
stub = dgraph.create_client_stub()
client = dgraph.create_client(stub)
dgraph.drop_all(client)
dgraph.add_schema(client)
dgraph.add_genome_to_schema(client, "test_genome")
for filepath in files.walkdir(path):
with open(filepath, 'rb') as file:
dgraph.create_graph(client, file, filepath)
assert se.type == SystemExit
def test_hash_verification():
path = os.path.abspath("pans_labyrinth/tests/data/contig_test")
stub = dgraph.create_client_stub()
client = dgraph.create_client(stub)
dgraph.drop_all(client)
dgraph.add_schema(client)
for filepath in files.walkdir(path):
with open(filepath, 'rb') as file:
x = 0
filename = file.name
genome = "genome_" + commandline.compute_hash(filepath)
dgraph.add_genome_to_schema(client, genome)
all_kmers = dgraph.get_kmers_files(filename, 11)
dgraph.add_kmers_dgraph(client, all_kmers, genome)
assert genome == "genome_" + commandline.compute_hash(filepath)
def test_verify_contig():
"""
A test to verify that the kmers returned from the graph are the same as the
kmers in the original contig
"""
#Create graph and fill it with kmers from a fasta file
path = os.path.abspath("pans_labyrinth/tests/data/contig_test")
stub = dgraph.create_client_stub()
client = dgraph.create_client(stub)
dgraph.drop_all(client)
dgraph.add_schema(client)
for filepath in files.walkdir(path):
with open(filepath, 'rb') as file:
x = 0
filename = file.name
genome = "genome_" + commandline.compute_hash(filepath)
dgraph.add_genome_to_schema(client, genome)
all_kmers = dgraph.get_kmers_files(filename, 11)
dgraph.add_kmers_dgraph(client, all_kmers, genome)
#Query graph and put all kmers into a list
sg1 = dgraph.path_query(client, genome)
kmer_list = []
for i, x in enumerate(sg1["path"]):
kmer = sg1["path"][i]["kmer"]
kmer_list.append(kmer)
print(kmer_list)
#Gets the whole first kmer in the list and then the last character of the rest of the kmers in the list
#Creates a string representing the contig
first, *rest = kmer_list
ends = [kmer[-1] for kmer in rest]
contig = ''.join([first] + ends)
#Gets the actual contig from the fasta file and turns it into string
with open(path + "/test.fasta", "r") as f:
for record in SeqIO.parse(f, "fasta"):
sequence = record.seq
sequence_string = str(sequence)
assert contig == sequence_string[:-1]