def test_distance_pep_google(self):
vector = bbc.create_vectors(self.pep_records, 10, utils.ALPHABET_PEP)
dist = bbc.Distance(vector, 'google')
matrix = distmatrix.create(self.pep_records.id_list, dist)
exp = [
" 4", "seq1 0.0000000 862.9400129 912.1886801 233.2058119",
"seq2 862.9400129 0.0000000 694.9729647 237.9720958",
"seq3 912.1886801 694.9729647 0.0000000 225.2262758",
"seq4 233.2058119 237.9720958 225.2262758 0.0000000"
]
self.assertEqual(matrix.format(), "\n".join(exp))
def main():
parser = get_parser()
args = validate_args(parser)
seq_records = seqrecords.read_fasta(args.fasta)
vector = bbc.create_vectors(seq_records, args.k, alphabet=args.alphabet)
dist = bbc.Distance(vector)
matrix = distmatrix.create(seq_records.id_list, dist)
if args.out:
oh = open(args.out, 'w')
matrix.write_to_file(oh, args.outfmt)
oh.close()
else:
matrix.display(args.outfmt)
if len(sys.argv) > 4:
sys.stderr.write("Too many arguments\n")
exit(1)
else:
method = sys.argv[1]
infile = sys.argv[2]
outfile = sys.argv[3]
input_file = open(infile, 'r')
seq_records = seqrecords.read_fasta(input_file)
input_file.close()
#choose one method to compute distance matrix
if method == "bbc":
alphabet = seqcontent.get_alphabet('protein')
vector = bbc.create_vectors(seq_records, 10, alphabet)
dist = bbc.Distance(vector)
matrix = distmatrix.create(seq_records.id_list, dist)
matrix.display()
elif method == "ncd":
dist = ncd.Distance(seq_records)
matrix = distmatrix.create(seq_records.id_list, dist)
matrix.display()
elif method == "wmetric":
matrix = subsmat.get('blosum62')
dist = wmetric.Distance(seq_records, matrix)
matrix = distmatrix.create(seq_records.id_list, dist)
matrix.display()
def test_create_vectors_on_pep_k1(self):
vec = bbc.create_vectors(self.pep_records, 1, utils.ALPHABET_PEP)
md5 = utils.calc_md5(vec)
self.assertEqual(vec.shape, (4, 400))
self.assertEqual(md5, 'd901d93d0d71102d0727633a67fc14a3')
def test_create_vectors_on_dna_k1(self):
vec = bbc.create_vectors(self.dna_records, 1, utils.ALPHABET_DNA)
md5 = utils.calc_md5(vec)
self.assertEqual(vec.shape, (3, 16))
self.assertEqual(md5, 'e99bc40356b00e04fd858a665af597ec')
def __init__(self, *args, **kwargs):
super(DistanceTest, self).__init__(*args, **kwargs)
utils.ModulesCommonTest.set_test_data()
self.vector = bbc.create_vectors(self.dna_records, 10,
utils.ALPHABET_DNA)
def test_create_vectors_throws_exception(self):
with self.assertRaises(Exception) as context:
bbc.create_vectors(self.dna_records, 60, utils.ALPHABET_DNA)
self.assertIn('Sequence too short', str(context.exception))