def test_passive_fit_window_escape():
n_points = 500
grand_up = np.ones(n_points)
grand_down = -grand_up
escape = 300
grand_down[escape:] = grand_down[escape:] * 2
t = np.arange(n_points)
assert preprocess.passive_fit_window(grand_up, grand_down, t) >= escape
assert preprocess.passive_fit_window(grand_up, grand_down, t) < t[-1]
def test_passive_fit_window_transient_after_start():
n_points = 500
grand_up = np.ones(n_points)
grand_down = -grand_up
escape = 300
grand_down[110] = grand_down[110] * 100
t = np.arange(n_points)
assert preprocess.passive_fit_window(
grand_up, grand_down, t, start_time=100) < t[-1]
def test_passive_fit_window_full_window():
n_points = 500
grand_up = np.ones(n_points)
grand_down = -grand_up
t = np.arange(n_points)
assert preprocess.passive_fit_window(grand_up, grand_down, t) == t[-1]
def main(paths, sweeps, dendrite_type, bridge_avg, passive_fit_start_time,
electrode_capacitance, junction_potential, random_seeds,
output_json, **kwargs):
"""Main sequence of pre-processing and passive fitting"""
# Extract Sweep objects (from IPFX package) from NWB file
nwb_path = paths["nwb"] # nwb - neurodata without borders (ephys data)
nwb_data = create_nwb_reader(nwb_path)
core_1_lsq, c1_start, c1_end = sweeps_from_nwb(
nwb_data, sweeps["core_1_long_squares"])
core_2_lsq, c2_start, c2_end = sweeps_from_nwb(
nwb_data, sweeps["core_2_long_squares"])
# Choose sweeps to train the model
sweep_set_to_fit, start, end = preprocess.select_core_1_or_core_2_sweeps(
core_1_lsq, c1_start, c1_end,
core_2_lsq, c2_start, c2_end)
if len(sweep_set_to_fit.sweeps) == 0:
ju.write(output_json, { 'error': "No usable sweeps found" })
return
# Calculate the target features from the training sweeps
step_analysis = StepAnalysis(start, end)
step_analysis.analyze(sweep_set_to_fit)
target_info = preprocess.target_features(
step_analysis.sweep_features(),
step_analysis.spikes_data())
stim_amp = step_analysis.sweep_features()["stim_amp"].values[0]
stim_dur = end - start
v_baseline = target_info.at["v_baseline", "mean"]
# Determine maximum current used for depolarization block checks
# during optimization
# Load noise sweeps to check highest current used
noise_1, _, _ = sweeps_from_nwb(
nwb_data, sweeps["seed_1_noise"])
noise_2, _, _ = sweeps_from_nwb(
nwb_data, sweeps["seed_2_noise"])
max_i = preprocess.max_i_for_depol_block_check(
core_1_lsq, core_2_lsq, noise_1, noise_2)
# Prepare inputs for passive fitting
is_spiny = dendrite_type == "spiny"
cap_checks, _, _ = sweeps_from_nwb(
nwb_data, sweeps["cap_checks"])
if len(cap_checks.sweeps) == 0:
logging.info("No cap check traces found")
should_run_passive_fit = False
passive_info = {
"should_run": False,
}
else:
grand_up, grand_down, t = preprocess.cap_check_grand_averages(cap_checks)
up_file, down_file = preprocess.save_grand_averages(
grand_up, grand_down, t, paths["storage_directory"])
escape_time = preprocess.passive_fit_window(grand_up, grand_down, t,
start_time=passive_fit_start_time)
passive_info = {
"should_run": True,
"bridge": bridge_avg,
"fit_window_start": passive_fit_start_time,
"fit_window_end": escape_time,
"electrode_cap": electrode_capacitance,
"is_spiny": is_spiny,
}
paths["up"] = up_file
paths["down"] = down_file
passive_info_path = os.path.join(
paths["storage_directory"], "passive_info.json")
ju.write(passive_info_path, passive_info)
# Determine whether morphology has an apical dendrite
has_apical = preprocess.swc_has_apical_compartments(paths["swc"])
# Decide which fits to run based on morphology and AP width
fit_types = preprocess.FitStyle.get_fit_types(
has_apical=has_apical,
is_spiny=is_spiny,
width=target_info.at["width", "mean"])
stage_1_tasks = [{"fit_type": fit_type, "seed": seed}
for seed in random_seeds
for fit_type in fit_types]
stage_2_tasks = [{"fit_type": preprocess.FitStyle.map_stage_2(fit_type), "seed": seed}
for seed in random_seeds
for fit_type in fit_types]
preprocess_results_path = os.path.join(
paths["storage_directory"], "preprocess_results.json")
ju.write(preprocess_results_path, {
"is_spiny": is_spiny,
"has_apical": has_apical,
"junction_potential": junction_potential,
"max_stim_test_na": max_i,
"v_baseline": v_baseline,
"stimulus": {
"amplitude": 1e-3 * stim_amp, # to nA
"delay": 1e3,
"duration": 1e3 * stim_dur, # to ms
},
"target_features": target_info.to_dict(orient="index"),
"sweeps": sweeps,
"sweeps_to_fit": [s.sweep_number for s in sweep_set_to_fit.sweeps],
})
paths.update({
"preprocess_results": preprocess_results_path,
"passive_info": passive_info_path,
})
output = {
"paths": paths,
"stage_1_task_list": stage_1_tasks,
"stage_2_task_list": stage_2_tasks,
}
ju.write(module.args["output_json"], output)