def t_ensembl_locus():
line = "LOCUS HG531_PATCH 1000000 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG531_PATCH", c.data.name
assert c._expected_size == 1000000, c._expected_size
line = "LOCUS HG531_PATCH 759984 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG531_PATCH", c.data.name
assert c._expected_size == 759984, c._expected_size
line = "LOCUS HG506_HG1000_1_PATCH 814959 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG506_HG1000_1_PATCH", c.data.name
assert c._expected_size == 814959, c._expected_size
line = "LOCUS HG506_HG1000_1_PATCH 1219964 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG506_HG1000_1_PATCH", c.data.name
assert c._expected_size == 1219964, c._expected_size
print("Done")
def t_ensembl_locus():
line = "LOCUS HG531_PATCH 1000000 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG531_PATCH", c.data.name
assert c._expected_size == 1000000, c._expected_size
line = "LOCUS HG531_PATCH 759984 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG531_PATCH", c.data.name
assert c._expected_size == 759984, c._expected_size
line = "LOCUS HG506_HG1000_1_PATCH 814959 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG506_HG1000_1_PATCH", c.data.name
assert c._expected_size == 814959, c._expected_size
line = "LOCUS HG506_HG1000_1_PATCH 1219964 bp DNA HTG 18-JUN-2011\n"
s = GenBank.Scanner.GenBankScanner()
c = GenBank._FeatureConsumer(True)
s._feed_first_line(c, line)
assert c.data.name == "HG506_HG1000_1_PATCH", c.data.name
assert c._expected_size == 1219964, c._expected_size
print("Done")
def test_topology_genbank(self):
"""Check GenBank LOCUS line parsing."""
# This is a bit low level, but can test pasing the LOCUS line only
tests = [
("LOCUS U00096",
None, None, None),
# This example is actually fungal, accession U49845 from Saccharomyces cerevisiae:
("LOCUS SCU49845 5028 bp DNA PLN 21-JUN-1999",
None, "DNA", "PLN"),
("LOCUS AB070938 6497 bp DNA linear BCT 11-OCT-2001",
"linear", "DNA", "BCT"),
("LOCUS NC_005816 9609 bp DNA circular BCT 21-JUL-2008",
"circular", "DNA", "BCT"),
("LOCUS SCX3_BUTOC 64 aa linear INV 16-OCT-2001",
"linear", None, "INV"),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner.GenBankScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(mt, mol_type,
"Wrong molecule_type %r not %r from %r" %
(mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(d, div,
"Wrong division %r not %r from %r" % (d, div, line))
def test_topology_genbank(self):
"""Check GenBank LOCUS line parsing."""
# This is a bit low level, but can test pasing the LOCUS line only
tests = [
("LOCUS U00096", None, None, None),
# This example is actually fungal, accession U49845 from Saccharomyces cerevisiae:
("LOCUS SCU49845 5028 bp DNA PLN 21-JUN-1999",
None, "DNA", "PLN"),
("LOCUS AB070938 6497 bp DNA linear BCT 11-OCT-2001",
"linear", "DNA", "BCT"),
("LOCUS NC_005816 9609 bp DNA circular BCT 21-JUL-2008",
"circular", "DNA", "BCT"),
("LOCUS SCX3_BUTOC 64 aa linear INV 16-OCT-2001",
"linear", None, "INV"),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner.GenBankScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(
t, topo, "Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(
mt, mol_type,
"Wrong molecule_type %r not %r from %r" % (mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(
d, div, "Wrong division %r not %r from %r" % (d, div, line))
def test_topology_genbank(self):
"""Check GenBank topology parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
("LOCUS U00096", None),
("LOCUS SCU49845 5028 bp DNA PLN 21-JUN-1999", None),
("LOCUS AB070938 6497 bp DNA linear BCT 11-OCT-2001", "linear"),
("LOCUS NC_005816 9609 bp DNA circular BCT 21-JUL-2008", "circular"),
("LOCUS SCX3_BUTOC 64 aa linear INV 16-OCT-2001", "linear"),
]
for (line, topo) in tests:
scanner = Scanner.GenBankScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
def test_topology_embl(self):
"""Check EMBL topology parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
("ID X56734; SV 1; linear; mRNA; STD; PLN; 1859 BP.", "linear"),
("ID CD789012; SV 4; linear; genomic DNA; HTG; MAM; 500 BP.", "linear"),
("ID BSUB9999 standard; circular DNA; PRO; 4214630 BP.", "circular"),
("ID SC10H5 standard; DNA; PRO; 4870 BP.", None),
("ID NRP_AX000635; PRT; NR1; 15 SQ", None),
("ID NRP0000016E; PRT; NR2; 5 SQ", None),
]
for (line, topo) in tests:
scanner = Scanner.EmblScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
def test_topology_genbank(self):
"""Check GenBank LOCUS line parsing."""
# This is a bit low level, but can test pasing the LOCUS line only
tests = [
("LOCUS U00096",
None, None, None, None),
# This example is actually fungal, accession U49845 from Saccharomyces cerevisiae:
("LOCUS SCU49845 5028 bp DNA PLN 21-JUN-1999",
None, "DNA", "PLN", None),
("LOCUS AB070938 6497 bp DNA linear BCT 11-OCT-2001",
"linear", "DNA", "BCT", None),
("LOCUS NC_005816 9609 bp DNA circular BCT 21-JUL-2008",
"circular", "DNA", "BCT", None),
("LOCUS SCX3_BUTOC 64 aa linear INV 16-OCT-2001",
"linear", None, "INV", None),
("LOCUS pEH010 5743 bp DNA circular",
"circular", "DNA", None, [BiopythonParserWarning]),
# This is a test of the format > 80 chars long
("LOCUS AZZZAA02123456789 1000000000 bp DNA linear PRI 15-OCT-2018",
"linear", "DNA", "PRI", None)
]
for (line, topo, mol_type, div, warning_list) in tests:
with warnings.catch_warnings(record=True) as caught:
warnings.simplefilter("always")
scanner = Scanner.GenBankScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(mt, mol_type,
"Wrong molecule_type %r not %r from %r" %
(mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(d, div,
"Wrong division %r not %r from %r" % (d, div, line))
if warning_list is None:
self.assertEqual(len(caught), 0)
else:
self.assertEqual(len(caught), len(warning_list))
for i, warning_class in enumerate(warning_list):
self.assertEqual(caught[i].category, warning_class)
def test_topology_embl(self):
"""Check EMBL ID line parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
# Modern examples with sequence version
("ID X56734; SV 1; linear; mRNA; STD; PLN; 1859 BP.", "linear",
"mRNA", "PLN"),
("ID CD789012; SV 4; linear; genomic DNA; HTG; MAM; 500 BP.",
"linear", "genomic DNA", "MAM"),
# Example to match GenBank example used above:
("ID U49845; SV 1; linear; genomic DNA; STD; FUN; 5028 BP.",
"linear", "genomic DNA", "FUN"),
# Old examples:
("ID BSUB9999 standard; circular DNA; PRO; 4214630 BP.",
"circular", "DNA", "PRO"),
("ID SC10H5 standard; DNA; PRO; 4870 BP.", None, "DNA", "PRO"),
# Patent example from 2016-06-10
# ftp://ftp.ebi.ac.uk/pub/databases/embl/patent/
("ID A01679; SV 1; linear; unassigned DNA; PAT; MUS; 12 BP.",
"linear", "unassigned DNA", "MUS"),
# Old patent examples
("ID NRP_AX000635; PRT; NR1; 15 SQ", None, None, "NR1"),
("ID NRP0000016E; PRT; NR2; 5 SQ", None, None, "NR2"),
# KIPO patent examples
("ID DI500001 STANDARD; PRT; 111 AA.", None, None,
None),
("ID DI644510 standard; PRT; 1852 AA.", None, None, None),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner.EmblScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(
t, topo, "Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(
mt, mol_type,
"Wrong molecule_type %r not %r from %r" % (mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(
d, div, "Wrong division %r not %r from %r" % (d, div, line))
def test_topology_embl(self):
"""Check EMBL ID line parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
# Modern examples with sequence version
("ID X56734; SV 1; linear; mRNA; STD; PLN; 1859 BP.",
"linear", "mRNA", "PLN"),
("ID CD789012; SV 4; linear; genomic DNA; HTG; MAM; 500 BP.",
"linear", "genomic DNA", "MAM"),
# Example to match GenBank example used above:
("ID U49845; SV 1; linear; genomic DNA; STD; FUN; 5028 BP.",
"linear", "genomic DNA", "FUN"),
# Old examples:
("ID BSUB9999 standard; circular DNA; PRO; 4214630 BP.",
"circular", "DNA", "PRO"),
("ID SC10H5 standard; DNA; PRO; 4870 BP.",
None, "DNA", "PRO"),
# Patent example from 2016-06-10
# ftp://ftp.ebi.ac.uk/pub/databases/embl/patent/
("ID A01679; SV 1; linear; unassigned DNA; PAT; MUS; 12 BP.",
"linear", "unassigned DNA", "MUS"),
# Old patent examples
("ID NRP_AX000635; PRT; NR1; 15 SQ", None, None, "NR1"),
("ID NRP0000016E; PRT; NR2; 5 SQ", None, None, "NR2"),
# KIPO patent examples
("ID DI500001 STANDARD; PRT; 111 AA.", None, None, None),
("ID DI644510 standard; PRT; 1852 AA.", None, None, None),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner.EmblScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(mt, mol_type,
"Wrong molecule_type %r not %r from %r" %
(mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(d, div,
"Wrong division %r not %r from %r" % (d, div, line))
def test_first_line_imgt(self):
"""Check IMGT ID line parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
("ID HLA00001 standard; DNA; HUM; 3503 BP.", None, "DNA",
"HUM"),
("ID HLA00001; SV 1; standard; DNA; HUM; 3503 BP.", None, "DNA",
"HUM"),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner._ImgtScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(
t, topo, "Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(
mt, mol_type,
"Wrong molecule_type %r not %r from %r" % (mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(
d, div, "Wrong division %r not %r from %r" % (d, div, line))
def test_first_line_imgt(self):
"""Check IMGT ID line parsing."""
# This is a bit low level, but can test pasing the ID line only
tests = [
("ID HLA00001 standard; DNA; HUM; 3503 BP.",
None, "DNA", "HUM"),
("ID HLA00001; SV 1; standard; DNA; HUM; 3503 BP.",
None, "DNA", "HUM"),
]
for (line, topo, mol_type, div) in tests:
scanner = Scanner._ImgtScanner()
consumer = GenBank._FeatureConsumer(1, GenBank.FeatureValueCleaner)
scanner._feed_first_line(consumer, line)
t = consumer.data.annotations.get('topology', None)
self.assertEqual(t, topo,
"Wrong topology %r not %r from %r" % (t, topo, line))
mt = consumer.data.annotations.get('molecule_type', None)
self.assertEqual(mt, mol_type,
"Wrong molecule_type %r not %r from %r" %
(mt, mol_type, line))
d = consumer.data.annotations.get('data_file_division', None)
self.assertEqual(d, div,
"Wrong division %r not %r from %r" % (d, div, line))
