def apply_feature_filters_length_test():
"""Test length filter function."""
# capture -- mean: 1; stdv: 0; median: 1; min: 1; max: 1; len: 6
capture = [1, 1, 1, 1, 1, 1]
# Only length filter -- pass (edge case, inclusive high)
filters = [filtering.LengthFilter(0, 6)]
pass_filters = filtering.apply_feature_filters(capture, filters)
assert pass_filters
# Only length filter -- pass (edge case, inclusive low)
filters = [filtering.LengthFilter(6, 10)]
pass_filters = filtering.apply_feature_filters(capture, filters)
assert pass_filters
# Only length filter -- fail (too short)
filters = [filtering.LengthFilter(8, 10)]
pass_filters = filtering.apply_feature_filters(capture, filters)
assert not pass_filters
# Only length filter -- fail (too long)
filters = [filtering.LengthFilter(0, 5)]
pass_filters = filtering.apply_feature_filters(capture, filters)
assert not pass_filters
# Only length filter -- pass (no filter actually given)
filters = [filtering.LengthFilter(None, None)]
pass_filters = filtering.apply_feature_filters(capture, filters)
assert pass_filters
def find_captures_4_unfolded_terminal_test():
"""Example capture window contains 1 long terminal capturethat has unfolded.
Tests: find_captures returns 1 capture
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_4.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(5_511_887, 5_604_585)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = True
filters = [filtering.LengthFilter(100, None)]
delay = 0
end_tol = 0
channel_number = 2
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 1
def find_captures_3_multicapture_nonterminal_test():
"""Example capture window contains 1 long terminal capture & 1 medium capture.
Tests: find_captures returns...
3 captures when terminal_capture_only = False
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_3.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(1_187_841, 1_280_674)
actual_captures = [(1_200_088, 1_201_033, False), (1_252_611, 1_280_674, True)]
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 3
end_tol = 0
channel_number = 2
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == len(actual_captures)
for test_capture in captures:
test_start = test_capture.window.start
test_end = test_capture.window.end
ejected = test_capture.ejected
assert (test_start, test_end, ejected) in actual_captures
def find_captures_3_multicapture_terminal_test():
"""Example capture window contains 1 long terminal capture & 2 medium/short captures.
Tests: find_captures returns...
1 capture when terminal_capture_only = True
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_3.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(1_187_841, 1_280_674)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = True
filters = [filtering.LengthFilter(100, None)]
delay = 0
end_tol = 0
channel_number = 2
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 1
def find_captures_2_nocaptures_test():
"""Example capture window contains no captures.
Test: find_captures returns no captures"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_2.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(3_423_474, 3_516_439)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 10
end_tol = 0
channel_number = 1
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 0
def find_captures_1_double_capture_noterminal_2_test():
"""Example capture window contains 2 long captures, neither terminal.
Also contains a few short blips.
Test: terminal_capture_only = False returns 2 captures"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_1.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(4_765_695, 4_858_482)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 0
end_tol = 0
channel_number = 1
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 2
def find_captures_0_single_capture_terminal_test():
data_file = "src/tests/data/capture_windows/test_data_capture_window_0.txt.gz"
window = Window(3_572_989, 3_665_680)
data = picoampere_signal_from_data_file(data_file)
actual_captures = [(33822 + window.start, 92691 + window.start, True)]
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = True
filters = [filtering.LengthFilter(100, None)]
delay = 0
end_tol = 0
channel_number = 1
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == len(actual_captures)
for test_capture in captures:
test_start = test_capture.window.start
test_end = test_capture.window.end
ejected = test_capture.ejected
assert (test_start, test_end, ejected) in actual_captures
def segment_test(self):
bulk_f5_fname = "src/tests/data/bulk_fast5_dummy.fast5"
filters = [filtering.LengthFilter(100, None)]
config = GeneralConfiguration(config={"n_workers": 2, "capture_directory": "src/tests"})
segment_config = {
"voltage_threshold": -180,
"signal_threshold_frac": 0.7,
"translocation_delay": 20,
"open_channel_prior_mean": 220,
"open_channel_prior_stdv": 50,
"good_channels": [
1,
2,
3,
], # this will be internally overwritten by the good channels calculation, which should not include channel 2
"end_tolerance": 50,
"terminal_capture_only": False,
"n_captures_per_file": 1000,
"bulkfast5": bulk_f5_fname,
}
segment_config = SegmentConfiguration(segment_config)
segment.segment(bulk_f5_fname, config, segment_config, overwrite=True, filters=filters)
run_id = "d0befb838f5a9a966e3c559dc3a75a6612745849"
actual_n_captures = 5
n_captures = 0
capture_f5_fname = f"src/tests/{run_id}_1.fast5"
with h5py.File(capture_f5_fname, "r") as f5:
for grp in f5.get("/"):
if "read" not in grp:
continue
n_captures += 1
d = f5[grp]
a = d["Signal"].attrs
start_time_local = a.get("start_time_local")
start_time_bulk = a.get("start_time_bulk")
assert start_time_local == start_time_bulk # No offset here
duration = a.get("duration")
len_signal = len(d["Signal"][()])
assert len_signal == duration
voltage = a.get("voltage")
assert voltage == segment_config.voltage_threshold
print(duration, a.get("channel_number"))
assert n_captures == actual_n_captures
os.remove(capture_f5_fname)
def parallel_find_captures_overflow_file_test(self):
bulk_f5_fname = "src/tests/data/bulk_fast5_dummy.fast5"
filters = [filtering.LengthFilter(100, None)]
config = GeneralConfiguration(config={"n_workers": 2, "capture_directory": "src/tests"})
segment_config = {
"voltage_threshold": -180,
"signal_threshold_frac": 0.7,
"translocation_delay": 20,
"open_channel_prior_mean": 220,
"open_channel_prior_stdv": 50,
"good_channels": [1, 3],
"end_tolerance": 50,
"terminal_capture_only": False,
"n_captures_per_file": 2,
"bulkfast5": bulk_f5_fname,
}
segment_config = SegmentConfiguration(segment_config)
segment.parallel_find_captures(config, segment_config, overwrite=True, filters=filters)
run_id = "d0befb838f5a9a966e3c559dc3a75a6612745849"
actual_n_captures = 5
n_captures = 0
capture_f5_fnames = [
os.path.join("src/tests/", x) for x in os.listdir("src/tests/") if run_id in x
]
assert len(capture_f5_fnames) == 3
for capture_f5_fname in capture_f5_fnames:
with h5py.File(capture_f5_fname, "r") as f5:
for grp in f5.get("/"):
if "read" not in grp:
continue
n_captures += 1
d = f5[grp]
a = d["Signal"].attrs
start_time_local = a.get("start_time_local")
start_time_bulk = a.get("start_time_bulk")
assert start_time_local == start_time_bulk # No offset here
duration = a.get("duration")
len_signal = len(d["Signal"][()])
assert len_signal == duration
voltage = a.get("voltage")
assert voltage == segment_config.voltage_threshold
print(duration, a.get("channel_number"))
os.remove(capture_f5_fname)
assert n_captures == actual_n_captures
def find_captures_7_capture_no_open_channel_test():
"""Example capture window contains 1 long terminal capture. Open pore region
is extremely, extremely short. Test by cutting off the open pore region.
Tests: find_captures returns 1 capture; open pore returns alt value.
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_7.txt.gz"
data = picoampere_signal_from_data_file(data_file)[100:]
window = Window(2_919_913, 3_013_723)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 100
end_tol = 0
channel_number = 2
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 1
# Rough check; should be ~229.05 & anything close is okay.
# The function is nondeterministic & should return this exact value, but if
# future changes are made, some tolerance can be allowed.
expected_open_channel_pA = 230
open_channel_pA = np.array([capture.open_channel_pA_calculated for capture in captures])
all_currents_within_bounds = all(
(np.isclose(open_channel_pA, expected_open_channel_pA, atol=0.5))
)
assert (
all_currents_within_bounds
), f"All captures should have calculated an open channel current close to {expected_open_channel_pA}."
def find_captures_5_unfolded_terminal_test():
"""Example capture window contains 1 long terminal capture. It was captured
almost immediately, causing a very short open pore region.
Tests: find_captures returns 1 capture
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_5.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(965_676, 1_059_216)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = True
filters = [filtering.LengthFilter(100, None)]
delay = 0
end_tol = 0
channel_number = 1
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 1
open_channel_pA = np.array([capture.open_channel_pA_calculated for capture in captures])
low_expected_open_channel_pA = 228.5
high_expected_open_channel_pA = 230
# Rough check; should be ~229.05 & anything close is okay.
# The function is nondeterministic & should return this exact value, but if
# future changes are made, some tolerance can be allowed.
all_currents_within_bounds = all(
(open_channel_pA > low_expected_open_channel_pA)
& (open_channel_pA < high_expected_open_channel_pA)
)
assert (
all_currents_within_bounds
), f"Expect all capture open channel currents to be between '{low_expected_open_channel_pA}' and '{high_expected_open_channel_pA}'."
def find_captures_8_capture_no_open_channel_test():
"""Example capture window contains 2 captures: both long, 1 terminal.
Test non-terminal long capture.
Tests: find_captures returns 2 captures.
Checks exact capture boundaries with delay = 3
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_8.txt.gz"
data = picoampere_signal_from_data_file(data_file)
window = Window(4_875_289, 4_969_337)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 3
end_tol = 0
channel_number = 2
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 2
actual_captures = [
(11310 + window.start, 22098 + window.start, False),
(26617 + window.start, 94048 + window.start, True),
]
for test_capture in captures:
test_start = test_capture.window.start
test_end = test_capture.window.end
ejected = test_capture.ejected
assert (test_start, test_end, ejected) in actual_captures
def find_captures_6_clog_no_open_channel_test():
"""Example capture window contains 1 long terminal capture. Open pore region
is extremely, extremely short. Test by cutting off the open pore region.
Tests: find_captures returns 1 capture; open pore returns alt value.
"""
data_file = "src/tests/data/capture_windows/test_data_capture_window_6.txt.gz"
data = picoampere_signal_from_data_file(data_file)[100:]
window = Window(2_769_436, 2_863_265)
signal_threshold_frac = 0.7
alt_open_channel_pA = 230
terminal_capture_only = False
filters = [filtering.LengthFilter(100, None)]
delay = 100
end_tol = 0
channel_number = 1
captures = segment.find_captures(
data,
channel_number,
window,
signal_threshold_frac,
alt_open_channel_pA,
terminal_capture_only=terminal_capture_only,
filters=filters,
delay=delay,
end_tol=end_tol,
)
assert len(captures) == 1
open_channel_pA = np.array([capture.open_channel_pA_calculated for capture in captures])
expected_open_channel_pA = 230
all_currents_within_bounds = all(
(np.isclose(open_channel_pA, expected_open_channel_pA, atol=0.5))
)
assert (
all_currents_within_bounds
), f"All calculated open channel currents should be close to {expected_open_channel_pA}"
