def test_swap(self):
'''test _swap'''
original_str = '\t'.join(
['1', '506', 'R', '300', '2', '500', 'L', '359'])
original_link = link.Link(None, None, None, s=original_str)
l = copy.copy(original_link)
expected_swap = link.Link(
None,
None,
None,
s='\t'.join(['2', '500', 'L', '359', '1', '506', 'R', '300']))
l._swap()
self.assertEqual(l, expected_swap)
l._swap()
self.assertEqual(l, original_link)
def test_update_from_sam(self):
'''test update_from_sam'''
sam_reader = pysam.Samfile(
os.path.join(data_dir, 'graph_test_update_from_sam.bam'), "rb")
ref_lengths = {}
pyfastaq.tasks.lengths_from_fai(
os.path.join(data_dir, 'graph_test_update_from_sam.ref.fa.fai'),
ref_lengths)
graph = scaffold_graph.Graph(ref_lengths)
for sam in sam_reader.fetch(until_eof=True):
graph.update_from_sam(sam, sam_reader)
self.assertEqual(len(graph.partial_links), 0)
self.assertEqual(len(graph.links), 1)
key = ('ref1', 'ref2')
self.assertTrue(key in graph.links)
expected_links = [
link.Link(None,
None,
None,
s='\t'.join([
'ref1', '506', 'R', '300', 'ref2', '500', 'L', '359'
]))
]
self.assertListEqual(graph.links[key], expected_links)
def test_merge(self):
'''test merge'''
link1 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '.']))
link2 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '.', '2', '500', 'L', '359']))
merged = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '359']))
with self.assertRaises(link.Error):
link1.merge(link1)
link2.merge(link2)
link1.merge(link2)
self.assertEqual(link1, merged)
def test_distance_to_contig_end(self):
'''test _distance_to_contig_end'''
links = [
link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '.'])),
link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '.', '2', '500', 'L', '359']))
]
self.assertEqual(links[0]._distance_to_contig_end(1), 205)
self.assertEqual(links[1]._distance_to_contig_end(2), 359)
with self.assertRaises(link.Error):
links[0]._distance_to_contig_end(2)
with self.assertRaises(link.Error):
links[1]._distance_to_contig_end(1)
def test_lt(self):
'''test __lt__'''
l = [
(['1', '10', 'R', '5', '2', '20', 'L',
'3'], ['1', '10', 'R', '5', '2', '20', 'L', '3'], False),
(['1', '10', 'R', '5', '2', '20', 'L',
'3'], ['1', '10', 'R', '5', '2', '20', 'L', '4'], True),
(['1', '10', 'R', '5', '2', '20', 'L',
'3'], ['1', '10', 'R', '6', '2', '20', 'L', '3'], True),
(['1', '10', 'R', '5', '2', '20', 'L',
'3'], ['2', '10', 'R', '6', '2', '20', 'L', '3'], True),
(['2', '10', 'R', '5', '2', '20', 'L',
'3'], ['1', '10', 'R', '6', '2', '20', 'L', '3'], False),
]
for t in l:
link1 = link.Link(None, None, None, '\t'.join(t[0]))
link2 = link.Link(None, None, None, '\t'.join(t[1]))
self.assertEqual(link1 < link2, t[2])
def test_insert_size(self):
'''test insert_size'''
link1 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '.']))
link2 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '.', '2', '500', 'L', '359']))
link3 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '359']))
with self.assertRaises(link.Error):
link1.insert_size()
with self.assertRaises(link.Error):
link2.insert_size()
self.assertEqual(link3.insert_size(), 564)
def test_distance_to_contig_ends(self):
'''test _distance_to_contig_ends'''
link1 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '.']))
link2 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '.', '2', '500', 'L', '359']))
link3 = link.Link(
None, None, None,
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '359']))
with self.assertRaises(link.Error):
link1._distance_to_contig_ends()
with self.assertRaises(link.Error):
link2._distance_to_contig_ends()
self.assertEqual(link3._distance_to_contig_ends(), (205, 359))
def test_init_no_links(self):
'''test link __init__ no links made'''
sam_reader = pysam.Samfile(
os.path.join(data_dir, 'link_test_init.reads.no_link.bam'), "rb")
ref_lengths = {}
pyfastaq.tasks.lengths_from_fai(
os.path.join(data_dir, 'link_test_init.ref.fa.fai'), ref_lengths)
links_from_bam = []
for sam in sam_reader.fetch(until_eof=True):
with self.assertRaises(link.Error):
l = link.Link(sam, sam_reader, ref_lengths)
def test_make_graph(self):
'''test _make_graph'''
ref_lengths = {'ref1': 100, 'ref2': 200, 'ref3': 300}
g = scaffold_graph.Graph(ref_lengths)
g.partial_links = {42: 42}
with self.assertRaises(scaffold_graph.Error):
g._make_graph(1000)
g.partial_links = {}
g.links[('ref1', 'ref2')] = [
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'R', '50', 'ref2', '200', 'L',
'10']))
]
g._make_graph(10)
self.assertEqual(len(g.contig_links), 0)
g._make_graph(1000)
self.assertEqual(len(g.contig_links), 1)
expected_contig_links = {('ref1', 'ref2'): {'RL': 1}}
self.assertDictEqual(g.contig_links, expected_contig_links)
g.links[('ref1', 'ref2')].append(
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'R', '50', 'ref2', '200', 'L',
'10'])))
g.links[('ref1', 'ref2')].append(
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'L', '50', 'ref2', '200', 'R',
'10'])))
g._make_graph(1000)
expected_contig_links = {('ref1', 'ref2'): {'RL': 2, 'LR': 1}}
self.assertDictEqual(g.contig_links, expected_contig_links)
def test_write_all_links_to_file(self):
'''test write_all_links_to_file'''
ref_lengths = {'ref1': 100, 'ref2': 200, 'ref3': 300}
g = scaffold_graph.Graph(ref_lengths)
g.links[('ref1', 'ref2')] = [
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'R', '50', 'ref2', '200', 'L',
'10']))
]
g.links[('ref1', 'ref2')].append(
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'R', '50', 'ref2', '200', 'L',
'10'])))
g.links[('ref1', 'ref2')].append(
link.Link(
None, None, None,
'\t'.join(['ref1', '100', 'L', '50', 'ref2', '200', 'R',
'10'])))
g.links[('ref3', 'ref4')] = [
link.Link(
None, None, None,
'\t'.join(['ref3', '100', 'R', '42', 'ref4', '200', 'L',
'42']))
]
tmp_file = 'tmp.contig_links'
g.write_all_links_to_file(tmp_file)
self.assertTrue(
filecmp.cmp(os.path.join(data_dir,
'graph_test_write_all_links_to_file.out'),
tmp_file,
shallow=False))
os.unlink(tmp_file)
def test_sort(self):
'''test sort'''
links = [
link.Link(None,
None,
None,
s='\t'.join([
'ref1', '500', 'L', '359', 'ref2', '506', 'R', '300'
])),
link.Link(None,
None,
None,
s='\t'.join([
'ref2', '500', 'L', '359', 'ref1', '506', 'R', '300'
]))
]
expected = [
link.Link(None,
None,
None,
s='\t'.join([
'ref1', '500', 'L', '359', 'ref2', '506', 'R', '300'
])),
link.Link(None,
None,
None,
s='\t'.join([
'ref1', '506', 'R', '300', 'ref2', '500', 'L', '359'
])),
]
assert len(links) == len(expected)
for i in range(len(links)):
links[i].sort()
self.assertEqual(links[i], expected[i])
def update_from_sam(self, sam, sam_reader):
'''Updates graph info from a pysam.AlignedSegment object'''
if sam.is_unmapped \
or sam.mate_is_unmapped \
or (sam.reference_id == sam.next_reference_id):
return
new_link = link.Link(sam, sam_reader, self.ref_lengths)
read_name = sam.query_name
if read_name in self.partial_links:
new_link.merge(self.partial_links[read_name])
del self.partial_links[read_name]
key = tuple(sorted((new_link.refnames[0], new_link.refnames[1])))
if key not in self.links:
self.links[key] = []
new_link.sort()
self.links[key].append(new_link)
else:
self.partial_links[read_name] = new_link
def test_init_with_link(self):
'''test link __init__ link made'''
sam_reader = pysam.Samfile(
os.path.join(data_dir, 'link_test_init.reads.make_link.bam'), "rb")
ref_lengths = {}
pyfastaq.tasks.lengths_from_fai(
os.path.join(data_dir, 'link_test_init.ref.fa.fai'), ref_lengths)
links_from_bam = []
for sam in sam_reader.fetch(until_eof=True):
links_from_bam.append(link.Link(sam, sam_reader, ref_lengths))
expected = [
'\t'.join(['1', '506', 'R', '300', '2', '500', 'L', '.']),
'\t'.join(['1', '506', 'R', '.', '2', '500', 'L', '359']),
]
self.assertEqual(len(expected), len(links_from_bam))
for i in range(len(expected)):
self.assertEqual(expected[i], str(links_from_bam[i]))
def test_parse(self):
'''test parse'''
ref_seqs = {}
ref_fasta = os.path.join(data_dir, 'bam_parse_test_parse.ref.fa')
bam = os.path.join(data_dir, 'bam_parse_test_parse.bam')
pyfastaq.tasks.file_to_dict(ref_fasta, ref_seqs)
bp = bam_parse.Parser(bam, ref_seqs)
bp.parse()
expected_soft_clipped = {'ref1': {2: [1, 0], 58: [0, 1]}}
expected_unmapped_mates = {'ref1': {2: 1}}
l = link.Link(
None,
None,
None,
s='\t'.join(['ref1', '500', 'R', '240', 'ref2', '500', 'L',
'299']))
expected_link_keys = [('ref1', 'ref2')]
self.assertEqual(expected_soft_clipped, bp.soft_clipped)
self.assertEqual(expected_unmapped_mates, bp.unmapped_mates)
self.assertEqual(list(bp.scaff_graph.links.keys()), expected_link_keys)
self.assertEqual(len(bp.scaff_graph.links[expected_link_keys[0]]), 1)
self.assertEqual(bp.scaff_graph.links[expected_link_keys[0]][0], l)