def test_iterate_cells_UMI_raise_unsorted(tmp_path):
"""
iterate_CB_UMI_of_busfile must raise an error of the busfile is unsorted (in terms of CBs)
"""
records = [
busio.Bus_record('ATAT', 'TAT', 14, 250, 13),
busio.Bus_record('ATAT', 'AAA', 10, 20, 1),
]
fname = tmp_path / 'some.bus'
busio.write_busfile(fname, records, cb_length=4, umi_length=3)
with pytest.raises(ValueError):
gen = pybustools.iterate_CB_UMI_of_busfile(fname)
list(gen)
"""
also check that it raises when the CB is unsorted
"""
records = [ # impotant: the UMI should be the same,
busio.Bus_record('TTAT', 'TAT', 14, 250, 13),
busio.Bus_record('ATAT', 'TAT', 10, 20, 1),
]
fname = tmp_path / 'some.bus'
busio.write_busfile(fname, records, cb_length=4, umi_length=3)
with pytest.raises(ValueError):
gen = pybustools.iterate_CB_UMI_of_busfile(fname)
list(gen)
def test_iterate_cb_umi(tmp_path):
records = [
busio.Bus_record('ATAT', 'AAA', 10, 20, 1),
busio.Bus_record('ATAT', 'GGG', 10, 20, 1),
busio.Bus_record('ATAT', 'GGG', 11, 20, 1),
busio.Bus_record('TAGA', 'TAT', 14, 250, 13),
busio.Bus_record('TTAT', 'AAA', 13, 206, 12),
]
fname = tmp_path / 'some.bus'
busio.write_busfile(fname, records, cb_length=4, umi_length=3)
gen = pybustools.iterate_CB_UMI_of_busfile(fname)
# the first record must have one entry
cb1, list1 = next(gen)
assert cb1 == ('ATAT', 'AAA') and len(list1) == 1
cb2, list2 = next(gen)
assert cb2 == ('ATAT', 'GGG') and len(list2) == 2
cb3, list3 = next(gen)
assert cb3 == ('TAGA', 'TAT') and len(list3) == 1
cb4, list4 = next(gen)
assert cb4 == ('TTAT', 'AAA') and len(list4) == 1
def subsample_bus_unseens_species(busfile, fractions):
"""
subsamples the given bus file at different depth, and records the number of reads/umis obtained
returns a DataFrame with the nReads and nUMIs as well as the UMI prevalences (duplicity of an umi -> #observations of that duplicity)
"""
gc.collect()
assert all([0 <= _ <= 1 for _ in fractions]), "Fractions must be in [0,1]"
reads = []
# lets group entires by CB/UMI instead of CB/UMI/EC (as done via B.iterate_bus())
# and sum the counts over ECs
for (cb, umi), record_list in tqdm.tqdm(iterate_CB_UMI_of_busfile(busfile, decode_seq=False)):
s = sum([r.COUNT for r in record_list])
reads.append(s)
reads = np.array(reads)
prevalences = collections.Counter(reads)
df = []
total = reads.sum()
for percent in fractions:
print(percent)
target = int(percent * total)
y = _downsample_array(reads, target=target, random_state=int(time.time()*1000), replace=False)
numis = (y > 0).sum()
df.append({
'nUMIs': numis,
'percent': percent,
'nReads': target,
'nReads2': y.sum()
})
df = pd.DataFrame(df)
return df, prevalences
def cell_umi_producer(fname, out_queue, decode_seq):
"""
turns iterate_CB_UMI_of_busfile into a queue
"""
for cb_umi, record_list in iterate_CB_UMI_of_busfile(fname, decode_seq):
# print('Putting record:', fname, record[0])
# due to some dumb pickling issues, we cant pickle a namedtuple directly
# hence deconstruct it and put that onto the queue
record_list = [_namedtuple_to_tuple(r) for r in record_list]
out_queue.put((cb_umi, record_list))
# print('sleeping')
# time.sleep(random.randrange(1,3))
out_queue.put(TERMINATOR)
def test_return_busrecord(tmp_path):
"""
make sure the functions return the namedtuple, not just the tuple
"""
records = [
busio.Bus_record('ATAT', 'AAA', 10, 20, 1),
busio.Bus_record('ATAT', 'GGG', 10, 20, 1),
busio.Bus_record('TAGA', 'TAT', 14, 250, 13),
busio.Bus_record('TTAT', 'AAA', 13, 206, 12),
]
fname = tmp_path / 'some.bus'
busio.write_busfile(fname, records, cb_length=4, umi_length=3)
for cb, record_list in pybustools.iterate_cells_of_busfile(fname):
for r in record_list:
assert isinstance(r, busio.Bus_record)
for cbumi, record_list in pybustools.iterate_CB_UMI_of_busfile(fname):
for r in record_list:
assert isinstance(r, busio.Bus_record)
def count_cb_umi_pairs(busfile):
counter = 0
for _ in tqdm.tqdm(iterate_CB_UMI_of_busfile(busfile, decode_seq=False)):
counter += 1
return counter