def test_sorted_items(self):
items = [1, 2, 3, 4, 4, 3, 2, 1]
unique_items = sorted_items(iter(items))
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
unique_items = sorted_items(iter(items), tempdir=tempfile.tempdir)
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
unique_items = sorted_items(iter(items), max_items_in_memory=3)
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
items = iter([])
assert not list(unique_items)
def test_key(self):
items = [(1, 'a'), (1, 'b'), (2, 'a')]
_sorted_items = sorted_items(iter(items), key=lambda x: x[0])
assert list(_sorted_items) == [(1, 'a'), (1, 'b'), (2, 'a')]
unique_items = unique(_sorted_items, key=lambda x: x[0])
assert list(unique_items) == [(1, 'a'), (2, 'a')]
_sorted_items = sorted_items(iter(items), key=lambda x: x[1])
assert list(_sorted_items) == [(1, 'a'), (2, 'a'), (1, 'b')]
unique_items = unique(_sorted_items, key=lambda x: x[1])
assert list(unique_items) == [(1, 'a'), (1, 'b')]
def test_key(self):
items = [(1, 'a'), (1, 'b'), (2, 'a')]
_sorted_items = list(sorted_items(iter(items), key=lambda x: x[0],
max_items_in_memory=1))
assert _sorted_items == [(1, 'a'), (1, 'b'), (2, 'a')]
unique_items = unique(_sorted_items, key=lambda x: x[0])
assert list(unique_items) == [(1, 'a'), (2, 'a')]
_sorted_items = sorted_items(iter(items), key=lambda x: x[1])
assert list(_sorted_items) == [(1, 'a'), (2, 'a'), (1, 'b')]
unique_items = unique(_sorted_items, key=lambda x: x[1])
assert list(unique_items) == [(1, 'a'), (1, 'b')]
def test_sorted_items(self):
items = [1, 2, 3, 4, 4, 3, 2, 1]
unique_items = sorted_items(iter(items))
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
unique_items = sorted_items(iter(items), tempdir=tempfile.tempdir)
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
unique_items = sorted_items(iter(items),
max_items_in_memory=3)
assert list(unique_items) == [1, 1, 2, 2, 3, 3, 4, 4]
items = iter([])
assert not list(unique_items)
def test_key(self):
items = [(1, 'a'), (1, 'b'), (2, 'a')]
_sorted_items = list(sorted_items(iter(items), key=lambda x: x[0],
max_items_in_memory=1))
assert _sorted_items == [(1, 'a'), (1, 'b'), (2, 'a')]
unique_items = unique(_sorted_items, key=lambda x: x[0])
assert list(unique_items) == [(1, 'a'), (2, 'a')]
_sorted_items = sorted_items(iter(items), key=lambda x: x[1])
assert list(_sorted_items) == [(1, 'a'), (2, 'a'), (1, 'b')]
unique_items = unique(_sorted_items, key=lambda x: x[1])
assert list(unique_items) == [(1, 'a'), (1, 'b')]
def sort_fastx_files(in_fhands, key, index_fpath=None, directory=None,
max_items_in_memory=None, tempdir=None):
if key == 'seq':
reads = read_seqs(in_fhands)
return sorted_items(reads, key=get_str_seq, tempdir=tempdir,
max_items_in_memory=max_items_in_memory)
elif key == 'coordinate':
return sort_by_position_in_ref(in_fhands, index_fpath=index_fpath,
directory=directory,
tempdir=tempdir)
elif key == 'name':
reads = read_seqs(in_fhands)
return sorted_items(reads, key=get_name, tempdir=tempdir,
max_items_in_memory=max_items_in_memory)
else:
raise ValueError('Non-supported sorting key')
def _get_paired_and_orphan(reads, ordered, max_reads_memory, temp_dir):
if ordered:
sorted_reads = reads
else:
def _key(seq):
return get_title(seq)
sorted_reads = sorted_items(reads, _key, max_reads_memory, temp_dir)
return group_pairs_by_name(sorted_reads)
def _get_paired_and_orphan(reads, ordered, max_reads_memory, temp_dir):
if ordered:
sorted_reads = reads
else:
def _key(seq):
return get_title(seq)
sorted_reads = sorted_items(reads, _key, max_reads_memory, temp_dir)
return group_pairs_by_name(sorted_reads)
def filter_duplicates(in_fhands, out_fhand, paired_reads,
n_seqs_packet=None, tempdir=None):
if not in_fhands:
raise ValueError('At least one input fhand is required')
pairs = _read_pairs(in_fhands, paired_reads)
sorted_pairs = sorted_items(pairs, key=_get_pair_key, tempdir=tempdir,
max_items_in_memory=n_seqs_packet)
for pair in unique(sorted_pairs, key=_get_pair_key):
write_seqs(pair, out_fhand)
def filter_duplicates(in_fhands, out_fhand, paired_reads, use_length=None,
n_seqs_packet=None, tempdir=None):
if not in_fhands:
raise ValueError('At least one input fhand is required')
pairs = _read_pairs(in_fhands, paired_reads)
get_pair_key = _PairKeyGetter(use_length=use_length)
if n_seqs_packet is None:
unique_pairs = unique_unordered(pairs, key=get_pair_key)
else:
sorted_pairs = sorted_items(pairs, key=get_pair_key, tempdir=tempdir,
max_items_in_memory=n_seqs_packet)
unique_pairs = unique(sorted_pairs, key=get_pair_key)
for pair in unique_pairs:
write_seqs(pair, out_fhand)
def filter_duplicates(in_fhands,
out_fhand,
paired_reads,
use_length=None,
n_seqs_packet=None,
tempdir=None):
if not in_fhands:
raise ValueError('At least one input fhand is required')
pairs = _read_pairs(in_fhands, paired_reads)
get_pair_key = _PairKeyGetter(use_length=use_length)
if n_seqs_packet is None:
unique_pairs = unique_unordered(pairs, key=get_pair_key)
else:
sorted_pairs = sorted_items(pairs,
key=get_pair_key,
tempdir=tempdir,
max_items_in_memory=n_seqs_packet)
unique_pairs = unique(sorted_pairs, key=get_pair_key)
for pair in unique_pairs:
write_seqs(pair, out_fhand)