def test_contigs_and_bases_that_hit_ref(self):
'''test _contigs_and_bases_that_hit_ref'''
self.qc.assembly_vs_ref_mummer_hits = {
'ctg1': [
mummer.NucmerHit('\t'.join(['1', '100', '1', '100', '100', '100', '100.00', '1008', '762', '1', '1', 'ref1', 'ctg1'])),
mummer.NucmerHit('\t'.join(['1', '100', '51', '150', '100', '100', '100.00', '1008', '762', '1', '1', 'ref1', 'ctg1'])),
],
'ctg2': [mummer.NucmerHit('\t'.join(['1', '42', '42', '84', '42', '84', '100.00', '42', '84', '1', '1', 'ref2', 'ctg2']))]
}
self.assertEqual((193, 2), self.qc._contigs_and_bases_that_hit_ref())
def test_get_unique_and_repetitive_from_contig_hits(self):
'''test _get_unique_and_repetitive_from_contig_hits'''
h1 = mummer.NucmerHit('\t'.join(['1', '10', '1', '10', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h2 = mummer.NucmerHit('\t'.join(['1', '10', '21', '30', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h3 = mummer.NucmerHit('\t'.join(['1', '10', '25', '35', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h4 = mummer.NucmerHit('\t'.join(['1', '10', '29', '40', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h5 = mummer.NucmerHit('\t'.join(['1', '10', '70', '90', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
hits = [h1, h2, h3, h4, h5]
expect_repetitive = [h2, h3, h4]
expect_unique = [h1, h5]
got_unique, got_repetitive = self.qc._get_unique_and_repetitive_from_contig_hits(hits)
self.assertEqual(expect_unique, got_unique)
self.assertEqual(expect_repetitive, got_repetitive)
def test_contig_placement_in_reference(self):
'''test _contig_placement_in_reference'''
h1 = mummer.NucmerHit('\t'.join(['1', '90', '100', '10', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h2 = mummer.NucmerHit('\t'.join(['17', '36', '21', '40', '100', '100', '100.00', '100', '100', '1', '+', 'ref2', 'qry1']))
expected = [(pyfastaq.intervals.Interval(9, 99), 'ref1', pyfastaq.intervals.Interval(0, 89), False, False)]
self.assertEqual(self.qc._contig_placement_in_reference([h1]), expected)
expected = [
(pyfastaq.intervals.Interval(9, 99), 'ref1', pyfastaq.intervals.Interval(0, 89), False, True),
(pyfastaq.intervals.Interval(20, 39), 'ref2', pyfastaq.intervals.Interval(16, 35), True, True)
]
self.assertEqual(self.qc._contig_placement_in_reference([h1, h2]), expected)
def test_mummer_coords_file_to_dict(self):
'''test _mummer_coords_file_to_dict'''
expected = {
'qry1': [
mummer.NucmerHit('\t'.join(['1', '100', '51', '150', '100', '100', '100.00', '1008', '762', '1', '1', 'ref1', 'qry1'])),
mummer.NucmerHit('\t'.join(['300', '500', '351', '550', '100', '100', '100.00', '1008', '762', '1', '1', 'ref2', 'qry1']))
],
'qry2': [mummer.NucmerHit('\t'.join(['1', '1000', '1', '1000', '1000', '1000', '100.00', '1000', '1542', '1', '1', 'ref2', 'qry2']))]
}
got = self.qc._mummer_coords_file_to_dict(os.path.join(data_dir, 'qc_test.mummer_coords_file_to_dict.coords'))
self.assertEqual(expected, got)
def test_hash_nucmer_hits_by_ref(self):
'''test _hash_nucmer_hits_by_ref'''
hit1 = mummer.NucmerHit('\t'.join(['1', '10', '20', '30', '10', '10', '100.00', '1008', '762', '1', '1', 'ref1', 'qry1']))
hit2 = mummer.NucmerHit('\t'.join(['1', '10', '20', '30', '10', '10', '100.00', '1008', '762', '1', '1', 'ref1', 'qry1']))
hit3 = mummer.NucmerHit('\t'.join(['1', '10', '20', '30', '10', '10', '100.00', '1008', '762', '1', '1', 'ref1', 'qry1']))
hit4 = mummer.NucmerHit('\t'.join(['1', '10', '20', '30', '10', '10', '100.00', '1008', '762', '1', '1', 'ref2', 'qry1']))
input_dict = {'x': [hit1, hit2, hit3, hit4]}
expected = {
'ref1': [hit1, hit2, hit3],
'ref2': [hit4]
}
got = self.qc._hash_nucmer_hits_by_ref(input_dict)
self.assertEqual(expected, got)
def test_ref_coords(self):
'''Test ref_coords'''
hits = ['\t'.join(['1', '100', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref']),
'\t'.join(['100', '1', '100', '1', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref'])
]
for h in hits:
m = mummer.NucmerHit(h)
self.assertEqual(pyfastaq.intervals.Interval(0,99), m.ref_coords())
def test_calculate_refseq_assembly_stats(self):
'''test _calculate_refseq_assembly_stats'''
self.qc.ref_ids = ['ref1', 'ref2', 'ref3', 'ref4']
self.qc.ref_lengths = {x:1000 for x in self.qc.ref_ids}
self.qc.assembly_vs_ref_mummer_hits = {'x': [
mummer.NucmerHit('\t'.join(['1', '1000', '1', '1000', '1000', '1000', '100.00', '1000', '1000', '1', '1', 'ref1', 'ctg1'])),
mummer.NucmerHit('\t'.join(['1', '800', '1', '800', '800', '800', '100.00', '800', '800', '1', '1', 'ref2', 'ctg2'])),
mummer.NucmerHit('\t'.join(['1', '500', '1', '500', '1000', '500', '100.00', '500', '500', '1', '1', 'ref3', 'ctg3.1'])),
mummer.NucmerHit('\t'.join(['501', '1000', '1', '500', '1000', '500', '100.00', '500', '500', '1', '1', 'ref3', 'ctg3.2']))
]
}
expected = {
'ref1': {
'hits': 1,
'bases_assembled': 1000,
'assembled': True,
'assembled_ok': True,
'longest_matching_contig': 1000,
},
'ref2': {
'hits': 1,
'bases_assembled': 800,
'assembled': False,
'assembled_ok': False,
'longest_matching_contig': 800,
},
'ref3': {
'hits': 2,
'bases_assembled': 1000,
'assembled': True,
'assembled_ok': False,
'longest_matching_contig': 500,
},
'ref4': {
'hits': 0,
'bases_assembled': 0,
'assembled': False,
'assembled_ok': False,
'longest_matching_contig': 0,
}
}
self.qc._calculate_refseq_assembly_stats()
self.maxDiff = None
self.assertEqual(expected, self.qc.refseq_assembly_stats)
def test_get_overlapping_qry_hits(self):
'''test _get_overlapping_qry_hits'''
h1 = mummer.NucmerHit('\t'.join(['1', '10', '1', '10', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h2 = mummer.NucmerHit('\t'.join(['1', '10', '21', '30', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h3 = mummer.NucmerHit('\t'.join(['1', '10', '25', '35', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h4 = mummer.NucmerHit('\t'.join(['1', '10', '29', '40', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
h5 = mummer.NucmerHit('\t'.join(['1', '10', '70', '90', '100', '100', '100.00', '100', '100', '1', '+', 'ref1', 'qry1']))
hits = [h1, h2, h3, h4, h5]
expected = [
[],
[h3, h4],
[h2, h4],
[h2, h3],
[]
]
self.assertEqual(len(hits), len(expected))
for i in range(len(hits)):
self.assertEqual(self.qc._get_overlapping_qry_hits(hits, hits[i]), expected[i])
def test_is_self_hit(self):
'''Test is_self_hit'''
tests = [('\t'.join(['1', '100', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref']), True),
('\t'.join(['1', '101', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref']), False),
('\t'.join(['2', '100', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref']), False),
('\t'.join(['1', '100', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref2']), False),
('\t'.join(['1', '100', '1', '100', '100', '100', '99.9', '1000', '1000', '1', '1', 'ref', 'ref']), False),
]
for t in tests:
m = mummer.NucmerHit(t[0])
self.assertEqual(m.is_self_hit(), t[1])
pass
def test_get_contig_hits_to_reference(self):
'''test _get_contig_hits_to_reference'''
self.qc.ref_fasta = os.path.join(data_dir, 'qc_test.reference.fa')
self.qc._set_ref_fa_data()
self.qc.ref_gff = os.path.join(data_dir, 'qc_test.reference.cds.gff')
self.qc.assembly_fasta = os.path.join(data_dir, 'qc_test.assembly.fasta')
self.qc._get_contig_hits_to_reference()
expected = {
'A:10-1017:+': [mummer.NucmerHit('\t'.join(['10', '1017', '1', '1008', '1008', '1008', '100.00', '1027', '1008', '1', '+', 'A', 'A:10-1017:+'])),
mummer.NucmerHit('\t'.join(['10', '240', '1', '231', '231', '231', '100.00', '240', '1008', '1', '+', 'A0', 'A:10-1017:+']))],
'B:1-1778:-': [mummer.NucmerHit('\t'.join(['1', '1778', '1778', '1', '1778', '1778', '100.00', '1778', '1778', '1', '+', 'B', 'B:1-1778:-']))],
'C:1-1413:+,E:1-200:-': [mummer.NucmerHit('\t'.join(['1', '1413', '1', '1413', '1413', '1413', '100.00', '1413', '1613', '1', '+', 'C', 'C:1-1413:+,E:1-200:-'])),
mummer.NucmerHit('\t'.join(['1', '200', '1414', '1613', '200', '200', '100.00', '890', '1613', '1', '+', 'E', 'C:1-1413:+,E:1-200:-']))],
'D:1-1000:+': [mummer.NucmerHit('\t'.join(['1', '1000', '1', '1000', '1000', '1000', '100.00', '1565', '1000', '1', '+', 'D', 'D:1-1000:+']))],
'E:400-700:-': [mummer.NucmerHit('\t'.join(['400', '700', '301', '1', '301', '301', '100.00', '890', '301', '1', '+', 'E', 'E:400-700:-']))],
'F:1-2341:+': [mummer.NucmerHit('\t'.join(['1', '2341', '1', '2341', '2341', '2341', '100.00', '2341', '2341', '1', '+', 'F', 'F:1-2341:+']))],
'F:1-2341:-': [mummer.NucmerHit('\t'.join(['1', '2341', '2341', '1', '2341', '2341', '100.00', '2341', '2341', '1', '+', 'F', 'F:1-2341:-']))]
}
self.assertDictEqual(expected, self.qc.assembly_vs_ref_mummer_hits)
def test_to_graph_edge(self):
'''Test to_graph_edge'''
hits = [
'\t'.join(['781', '981', '10', '210', '200', '200', '98', '1000', '1000', '1', '1', 'ref', 'qry']), # %id too low
'\t'.join(['781', '980', '10', '210', '199', '200', '98', '1000', '1000', '1', '1', 'ref', 'qry']), # hit too short
'\t'.join(['781', '981', '10', '209', '200', '199', '98', '1000', '1000', '1', '1', 'ref', 'qry']), # hit too short
'\t'.join(['1', '200', '1', '200', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # bad orientation
'\t'.join(['200', '1', '200', '1', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # bad orientation
'\t'.join(['800', '1000', '800', '1000', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # bad orientation
'\t'.join(['1000', '800', '1000', '800', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # bad orientation
'\t'.join(['300', '500', '300', '500', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # not at ends
'\t'.join(['1', '1000', '1', '1000', '1000', '1000', '100.00', '1000', '1000', '1', '1', 'ref', 'qry']), # whole contigs hit
'\t'.join(['1', '500', '1', '1000', '500', '500', '100.00', '500', '1000', '1', '1', 'ref1', 'qry1']), # contained contig
'\t'.join(['1', '200', '791', '991', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref2', 'qry2']),
'\t'.join(['781', '981', '10', '210', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref3', 'qry3']),
'\t'.join(['991', '791', '781', '981', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref4', 'qry4']),
'\t'.join(['210', '10', '15', '215', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref5', 'qry5']),
'\t'.join(['781', '981', '991', '791', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref6', 'qry6']),
'\t'.join(['10', '210', '215', '5', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref7', 'qry7']),
'\t'.join(['210', '10', '980', '780', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref8', 'qry8']),
'\t'.join(['995', '795', '215', '15', '200', '200', '100.00', '1000', '1000', '1', '1', 'ref9', 'qry9']),
]
expected = [
None,
None,
None,
None,
None,
None,
None,
None,
None,
None,
edge.Edge('qry2', 790, 990, 'ref2', 0, 199),
edge.Edge('ref3', 780, 980, 'qry3', 9, 209),
edge.Edge('qry4', 780, 980, 'ref4', 990, 790),
edge.Edge('ref5', 209, 9, 'qry5', 14, 214),
edge.Edge('ref6', 780, 980, 'qry6', 990, 790),
edge.Edge('qry7', 214, 4, 'ref7', 9, 209),
edge.Edge('ref8', 209, 9, 'qry8', 979, 779),
edge.Edge('qry9', 214, 14, 'ref9', 994, 794),
]
assert len(expected) == len(hits)
for i in range(len(hits)):
m = mummer.NucmerHit(hits[i])
self.assertEqual(m.to_graph_edge(), expected[i])
def test_is_at_ends(self):
'''Test is_at_ends'''
tests = [('\t'.join(['1', '100', '200', '300', '100', '100', '100.00', '1000', '500', '1', '1', 'ref', 'qry']), False, mummer.HIT_AT_START),
('\t'.join(['51', '151', '200', '300', '100', '100', '100.00', '1000', '500', '1', '1', 'ref', 'qry']), False, mummer.HIT_NO_ENDS),
('\t'.join(['900', '1000', '200', '300', '100', '100', '100.00', '1000', '500', '1', '1', 'ref', 'qry']), False, mummer.HIT_AT_END),
('\t'.join(['1000', '900', '200', '300', '100', '100', '100.00', '1000', '500', '1', '1', 'ref', 'qry']), False, mummer.HIT_AT_END),
('\t'.join(['850', '949', '200', '300', '100', '100', '100.00', '1000', '500', '1', '1', 'ref', 'qry']), False, mummer.HIT_NO_ENDS),
('\t'.join(['42', '992', '200', '1152', '950', '950', '100.00', '1000', '5000', '1', '1', 'ref', 'qry']), False, mummer.HIT_AT_BOTH_ENDS),
('\t'.join(['200', '300', '1', '100', '100', '100', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_AT_START),
('\t'.join(['200', '300', '51', '151', '100', '100', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_NO_ENDS),
('\t'.join(['200', '300', '900', '1000', '100', '100', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_AT_END),
('\t'.join(['200', '300', '1000', '900', '100', '100', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_AT_END),
('\t'.join(['200', '300', '850', '949', '100', '100', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_NO_ENDS),
('\t'.join(['200', '300', '42', '992', '950', '950', '100.00', '500', '1000', '1', '1', 'ref', 'qry']), True, mummer.HIT_AT_BOTH_ENDS),
]
for t in tests:
m = mummer.NucmerHit(t[0])
self.assertEqual(m._is_at_ends(use_qry=t[1]), t[2])
def test_on_same_strand(self):
'''test on_same_strand'''
self.assertTrue(mummer.NucmerHit('\t'.join(['1', '100', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref'])).on_same_strand())
self.assertTrue(mummer.NucmerHit('\t'.join(['100', '1', '100', '1', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref'])).on_same_strand())
self.assertFalse(mummer.NucmerHit('\t'.join(['1', '100', '100', '1', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref'])).on_same_strand())
self.assertFalse(mummer.NucmerHit('\t'.join(['100', '1', '1', '100', '100', '100', '100.00', '1000', '1000', '1', '1', 'ref', 'ref'])).on_same_strand())
