def match_pairs(reads, out_fhand, orphan_out_fhand, out_format, ordered=True,
check_order_buffer_size=0, max_reads_memory=None,
temp_dir=None):
'''It matches the seq pairs in an iterator and splits the orphan seqs.'''
counts = 0
check_order_buffer = KeyedSet()
for pair in _get_paired_and_orphan(reads, ordered, max_reads_memory,
temp_dir):
if len(pair) == 1:
write_seqs(pair, orphan_out_fhand, out_format)
try:
name = _parse_pair_direction_and_name(pair[0])[0]
except PairDirectionError:
name = get_name(pair[0])
if ordered and counts < check_order_buffer_size:
counts += 1
if not check_order_buffer.check_add(name):
msg = 'Reads are not ordered by pairs.Use unordered option'
raise ItemsNotSortedError(msg)
elif ordered and counts >= check_order_buffer_size:
if name in check_order_buffer:
msg = 'Reads are not ordered by pairs.Use unordered option'
raise ItemsNotSortedError(msg)
elif len(pair) == 2:
write_seqs(pair, out_fhand, out_format)
flush_fhand(orphan_out_fhand)
flush_fhand(out_fhand)
def match_pairs(reads,
out_fhand,
orphan_out_fhand,
out_format,
ordered=True,
check_order_buffer_size=0,
max_reads_memory=None,
temp_dir=None):
'''It matches the seq pairs in an iterator and splits the orphan seqs.'''
counts = 0
check_order_buffer = KeyedSet()
for pair in _get_paired_and_orphan(reads, ordered, max_reads_memory,
temp_dir):
if len(pair) == 1:
write_seqs(pair, orphan_out_fhand, out_format)
try:
name = _parse_pair_direction_and_name(pair[0])[0]
except PairDirectionError:
name = get_name(pair[0])
if ordered and counts < check_order_buffer_size:
counts += 1
if not check_order_buffer.check_add(name):
msg = 'Reads are not ordered by pairs.Use unordered option'
raise ItemsNotSortedError(msg)
elif ordered and counts >= check_order_buffer_size:
if name in check_order_buffer:
msg = 'Reads are not ordered by pairs.Use unordered option'
raise ItemsNotSortedError(msg)
elif len(pair) == 2:
write_seqs(pair, out_fhand, out_format)
flush_fhand(orphan_out_fhand)
flush_fhand(out_fhand)
def test_unordered_set(self):
in_set = [1, 2, 3, 4, 5, 8, 10]
not_in_set = [6, 9, 11, 13]
keyed_set = KeyedSet(in_set)
for item in in_set:
assert item in keyed_set
assert keyed_set.check_add(7)
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert not keyed_set.check_add(2)
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert keyed_set.check_add(0)
assert keyed_set._items == set([0, 1, 2, 3, 4, 5, 7, 8, 10])
for item in not_in_set:
assert item not in keyed_set
#with key
a = 'a'
in_set = [(1, a), (2, a), (3, a), (4, a), (5, a), (8, a), (10, a)]
not_in_set = [(6, a), (9, a), (11, a), (13, a)]
def key(item):
return item[0]
keyed_set = KeyedSet(in_set, key=key)
for item in in_set:
assert item in keyed_set
assert keyed_set.check_add((7, a))
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert not keyed_set.check_add((2, a))
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert keyed_set.check_add((0, a))
assert keyed_set._items == set([0, 1, 2, 3, 4, 5, 7, 8, 10])
for item in not_in_set:
assert item not in keyed_set
def test_unordered_set(self):
in_set = [1, 2, 3, 4, 5, 8, 10]
not_in_set = [6, 9, 11, 13]
keyed_set = KeyedSet(in_set)
for item in in_set:
assert item in keyed_set
assert keyed_set.check_add(7)
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert not keyed_set.check_add(2)
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert keyed_set.check_add(0)
assert keyed_set._items == set([0, 1, 2, 3, 4, 5, 7, 8, 10])
for item in not_in_set:
assert item not in keyed_set
#with key
a = 'a'
in_set = [(1, a), (2, a), (3, a), (4, a), (5, a), (8, a), (10, a)]
not_in_set = [(6, a), (9, a), (11, a), (13, a)]
def key(item):
return item[0]
keyed_set = KeyedSet(in_set, key=key)
for item in in_set:
assert item in keyed_set
assert keyed_set.check_add((7, a))
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert not keyed_set.check_add((2, a))
assert keyed_set._items == set([1, 2, 3, 4, 5, 7, 8, 10])
assert keyed_set.check_add((0, a))
assert keyed_set._items == set([0, 1, 2, 3, 4, 5, 7, 8, 10])
for item in not_in_set:
assert item not in keyed_set