class AlignmentCodonFilters(unittest.TestCase):
def setUp(self):
if not os.path.exists(temp_dir):
os.makedirs(temp_dir)
self.aln_obj = AlignmentList([], sql_db=sql_db)
def tearDown(self):
self.aln_obj.clear_alignments()
self.aln_obj.con.close()
shutil.rmtree(temp_dir)
def test_codon_filter_pos1(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, False, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["a" * 16] * 10)
def test_codon_filter_pos2(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([False, True, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["t" * 16] * 10)
def test_codon_filter_pos3(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([False, False, True],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["g" * 16] * 10)
def test_codon_filter_pos12(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, True, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["at" * 16] * 10)
def test_codon_filter_pos13(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, False, True],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["ag" * 16] * 10)
def test_codon_filter_all(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, True, True])
s = []
for _, seq, _ in self.aln_obj.iter_alignments():
s.append(seq)
self.assertEqual(s, ["atg" * 16] * 10)
class AlignmentCodonFilters(unittest.TestCase):
def setUp(self):
if not os.path.exists(temp_dir):
os.makedirs(temp_dir)
self.aln_obj = AlignmentList([],sql_db=sql_db)
def tearDown(self):
self.aln_obj.clear_alignments()
self.aln_obj.con.close()
shutil.rmtree(temp_dir)
def test_codon_filter_pos1(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, False, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["a" * 16] * 10)
def test_codon_filter_pos2(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([False, True, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["t" * 16] * 10)
def test_codon_filter_pos3(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([False, False, True],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["g" * 16] * 10)
def test_codon_filter_pos12(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, True, False],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["at" * 16] * 10)
def test_codon_filter_pos13(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, False, True],
table_out="master_out")
s = []
for _, seq, _ in self.aln_obj.iter_alignments("master_out"):
s.append(seq)
self.assertEqual(s, ["ag" * 16] * 10)
def test_codon_filter_all(self):
self.aln_obj.add_alignment_files(codon_filter)
self.aln_obj.filter_codon_positions([True, True, True])
s = []
for _, seq, _ in self.aln_obj.iter_alignments():
s.append(seq)
self.assertEqual(s, ["atg" * 16] * 10)
class LoadAlignmentsTest(unittest.TestCase):
def setUp(self):
if not os.path.exists(temp_dir):
os.makedirs(temp_dir)
self.aln_obj = AlignmentList([], sql_db=sql_db)
def tearDown(self):
self.aln_obj.clear_alignments()
self.aln_obj.con.close()
shutil.rmtree(temp_dir)
def test_dna_load(self):
self.aln_obj = AlignmentList(dna_data_fas, sql_db=sql_db)
self.assertEqual(["DNA"], self.aln_obj.sequence_code)
def test_protein_load(self):
self.aln_obj = AlignmentList(protein_no_missing, sql_db=sql_db)
self.assertEqual(["Protein"], self.aln_obj.sequence_code)
def test_mixed_type_load(self):
self.aln_obj = AlignmentList(mixed_seq_type, sql_db=sql_db)
self.assertEqual(["DNA", "Protein"],
sorted(self.aln_obj.sequence_code))
def test_class_instance(self):
self.aln_obj = AlignmentList([], sql_db=sql_db)
self.assertIsInstance(self.aln_obj.alignments, OrderedDict)
def test_load_fas(self):
self.aln_obj = AlignmentList(dna_data_fas, sql_db=sql_db)
def test_load_single_fas(self):
single_aln = Alignment(dna_data_fas[0], sql_cursor=self.aln_obj.cur)
def test_load_phy(self):
self.aln_obj = AlignmentList(dna_data_phy, sql_db=sql_db)
def test_load_single_phy(self):
single_aln = Alignment(dna_data_phy[0], sql_cursor=self.aln_obj.cur)
def test_load_single_interleave_phy(self):
single_aln = Alignment(phylip_interleave[0],
sql_cursor=self.aln_obj.cur)
def test_load_nex(self):
self.aln_obj = AlignmentList(dna_data_nex, sql_db=sql_db)
def test_load_single_nex(self):
single_aln = Alignment(dna_data_nex[0], sql_cursor=self.aln_obj.cur)
def test_load_interleave_nex(self):
single_aln = Alignment(concatenated_interleave_nexus[0],
sql_cursor=self.aln_obj.cur)
def test_load_stc(self):
self.aln_obj = AlignmentList(dna_data_stc, sql_db=sql_db)
def test_load_single_stc(self):
single_aln = Alignment(dna_data_stc[0], sql_cursor=self.aln_obj.cur,
db_idx=self.aln_obj._idx + 1)
def test_load_loci(self):
self.aln_obj = AlignmentList(dna_data_loci, sql_db=sql_db)
def test_load_single_loci(self):
single_aln = Alignment(dna_data_loci[0], sql_cursor=self.aln_obj.cur,
db_idx=self.aln_obj._idx + 1)
def test_load_nexus_par(self):
self.aln_obj = AlignmentList(concatenated_medium_nexus, sql_db=sql_db)
self.assertTrue(self.aln_obj.partitions.partitions)
def test_load_wrong_type(self):
self.aln_obj = AlignmentList(bad_file, sql_db=sql_db)
self.assertTrue(self.aln_obj.bad_alignments)
def test_duplicate_files(self):
self.aln_obj = AlignmentList(dna_data_loci + dna_data_loci,
sql_db=sql_db)
self.assertTrue(self.aln_obj.duplicate_alignments)
def test_unequal_length(self):
self.aln_obj = AlignmentList(unequal_file, sql_db=sql_db)
self.assertTrue(self.aln_obj.non_alignments)
def test_bad_file_removal_from_db(self):
self.aln_obj = AlignmentList(unequal_file, sql_db=sql_db)
self.aln_obj.add_alignment_files(dna_data_fas)
s = 0
for tx, _, _ in self.aln_obj.iter_alignments():
if tx == "Seq1":
s += 1
self.assertEqual(s, 0)
def test_load_no_data(self):
self.aln_obj = AlignmentList(no_data, sql_db=sql_db)
def test_alternative_missing(self):
self.aln_obj = AlignmentList(alternative_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "?")
def test_dna_missing_default(self):
self.aln_obj = AlignmentList(single_dna, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "n")
def test_protein_missing_default(self):
self.aln_obj = AlignmentList(protein_no_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "x")
def test_dna_missing_eval(self):
self.aln_obj = AlignmentList(concatenated_medium_nexus, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "n")
def test_protein_missing_eval(self):
self.aln_obj = AlignmentList(protein_normal_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "x")
def test_non_ascii_taxon_names(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
non_ascii_tx = [x for x in self.aln_obj.taxa_names if
x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><']
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_iteration(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
non_ascii_tx = []
for tx, _, _ in self.aln_obj.iter_alignments():
if tx == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><':
non_ascii_tx.append(tx)
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_get_taxaidx(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
non_ascii_tx = [x for x in aln.taxa_idx
if x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><']
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_iter_columns(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
tx_list, _, _ = next(self.aln_obj.iter_columns(include_taxa=True))
non_ascii_tx = [x for x in tx_list
if x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><']
self.assertEqual(len(non_ascii_tx), 1)
class LoadAlignmentsTest(unittest.TestCase):
def setUp(self):
if not os.path.exists(temp_dir):
os.makedirs(temp_dir)
self.aln_obj = AlignmentList([], sql_db=sql_db)
def tearDown(self):
self.aln_obj.clear_alignments()
self.aln_obj.con.close()
shutil.rmtree(temp_dir)
def test_dna_load(self):
self.aln_obj = AlignmentList(dna_data_fas, sql_db=sql_db)
self.assertEqual(["DNA"], self.aln_obj.sequence_code)
def test_protein_load(self):
self.aln_obj = AlignmentList(protein_no_missing, sql_db=sql_db)
self.assertEqual(["Protein"], self.aln_obj.sequence_code)
def test_mixed_type_load(self):
self.aln_obj = AlignmentList(mixed_seq_type, sql_db=sql_db)
self.assertEqual(["DNA", "Protein"],
sorted(self.aln_obj.sequence_code))
def test_class_instance(self):
self.aln_obj = AlignmentList([], sql_db=sql_db)
self.assertIsInstance(self.aln_obj.alignments, OrderedDict)
def test_load_fas(self):
self.aln_obj = AlignmentList(dna_data_fas, sql_db=sql_db)
def test_load_single_fas(self):
single_aln = Alignment(dna_data_fas[0], sql_cursor=self.aln_obj.cur)
def test_load_phy(self):
self.aln_obj = AlignmentList(dna_data_phy, sql_db=sql_db)
def test_load_single_phy(self):
single_aln = Alignment(dna_data_phy[0], sql_cursor=self.aln_obj.cur)
def test_load_single_interleave_phy(self):
single_aln = Alignment(phylip_interleave[0],
sql_cursor=self.aln_obj.cur)
def test_load_nex(self):
self.aln_obj = AlignmentList(dna_data_nex, sql_db=sql_db)
def test_load_single_nex(self):
single_aln = Alignment(dna_data_nex[0], sql_cursor=self.aln_obj.cur)
def test_load_interleave_nex(self):
single_aln = Alignment(concatenated_interleave_nexus[0],
sql_cursor=self.aln_obj.cur)
def test_load_stc(self):
self.aln_obj = AlignmentList(dna_data_stc, sql_db=sql_db)
def test_load_single_stc(self):
single_aln = Alignment(dna_data_stc[0],
sql_cursor=self.aln_obj.cur,
db_idx=self.aln_obj._idx + 1)
def test_load_loci(self):
self.aln_obj = AlignmentList(dna_data_loci, sql_db=sql_db)
def test_load_single_loci(self):
single_aln = Alignment(dna_data_loci[0],
sql_cursor=self.aln_obj.cur,
db_idx=self.aln_obj._idx + 1)
def test_load_nexus_par(self):
self.aln_obj = AlignmentList(concatenated_medium_nexus, sql_db=sql_db)
self.assertTrue(self.aln_obj.partitions.partitions)
def test_load_wrong_type(self):
self.aln_obj = AlignmentList(bad_file, sql_db=sql_db)
self.assertTrue(self.aln_obj.bad_alignments)
def test_duplicate_files(self):
self.aln_obj = AlignmentList(dna_data_loci + dna_data_loci,
sql_db=sql_db)
self.assertTrue(self.aln_obj.duplicate_alignments)
def test_unequal_length(self):
self.aln_obj = AlignmentList(unequal_file, sql_db=sql_db)
self.assertTrue(self.aln_obj.non_alignments)
def test_bad_file_removal_from_db(self):
self.aln_obj = AlignmentList(unequal_file, sql_db=sql_db)
self.aln_obj.add_alignment_files(dna_data_fas)
s = 0
for tx, _, _ in self.aln_obj.iter_alignments():
if tx == "Seq1":
s += 1
self.assertEqual(s, 0)
def test_load_no_data(self):
self.aln_obj = AlignmentList(no_data, sql_db=sql_db)
def test_alternative_missing(self):
self.aln_obj = AlignmentList(alternative_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "?")
def test_dna_missing_default(self):
self.aln_obj = AlignmentList(single_dna, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "n")
def test_protein_missing_default(self):
self.aln_obj = AlignmentList(protein_no_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "x")
def test_dna_missing_eval(self):
self.aln_obj = AlignmentList(concatenated_medium_nexus, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "n")
def test_protein_missing_eval(self):
self.aln_obj = AlignmentList(protein_normal_missing, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
self.assertEqual(aln.sequence_code[1], "x")
def test_non_ascii_taxon_names(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
non_ascii_tx = [
x for x in self.aln_obj.taxa_names
if x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><'
]
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_iteration(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
non_ascii_tx = []
for tx, _, _ in self.aln_obj.iter_alignments():
if tx == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><':
non_ascii_tx.append(tx)
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_get_taxaidx(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
aln = self.aln_obj.alignments.values()[0]
non_ascii_tx = [
x for x in aln.taxa_idx
if x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><'
]
self.assertEqual(len(non_ascii_tx), 1)
def test_non_ascii_iter_columns(self):
self.aln_obj = AlignmentList(non_ascii, sql_db=sql_db)
tx_list, _, _ = next(self.aln_obj.iter_columns(include_taxa=True))
non_ascii_tx = [
x for x in tx_list
if x == '\xc3\xa9!"#$%&/=?\'~\xc2\xba\xc2\xaa^"><'
]
self.assertEqual(len(non_ascii_tx), 1)
