def _test_FEATHER_retract_only(fec, threshold, tau):
"""
:param fec: the force-extension curve, from (e.g.)
_command_line_config.make_fec
:param threshold: the significance threshold
:param tau: the fractional smoothing
:return:
"""
# # Next, we show how to get do event detection based only on the approach
# We do the following steps(1-4), below
# (1) Create a new FEC based only on the retract
# first, we get just the retract, which starts at t=(TriggerTime+DwellTime)
retract = Analysis.zero_and_split_force_extension_curve(fec, tau).retract
# next, we use <pct_pseudo_approach> of the retract to make an approach
pct_pseudo_approach = 0.1
split = UtilFEATHER._split_from_retract(retract, pct_pseudo_approach, tau)
# combined the 'fake' approach and retract
to_save = FEC_Util._create_psuedo_fec_to_save(split)
to_save.Meta.SourceFile = "TemporaryFEC"
# (2) Save the FEC out to a csv
file_saved = FEC_Util._save_single_csv(out_dir="./", fec_to_save=to_save)
# (3) Run FEATHER on the CSV, just as we did above
time, sep, force, file_dict = _get_file_info(file_saved)
meta_dict = dict(threshold=threshold, tau=tau, **file_dict)
event_indices_1 = _command_line_config.run_feather(in_file=file_saved,
**meta_dict)
_plot_results(time, force, event_indices_1)
def _analyze_file(in_file):
threshold = 1e-3
tau = 2e-2
time, sep, force, file_dict = _get_file_info(in_file)
meta_dict = dict(threshold=threshold, tau=tau, **file_dict)
event_indices_1 = _command_line_config.run_feather(in_file=in_file,
**meta_dict)
# # (2) directly, using python arrays and a constructed fec object. This is
# # likely to be much faster, since there is no extra file IO.
fec = _command_line_config.make_fec(time=time,
separation=sep,
force=force,
**meta_dict)
event_indices_2 = _command_line_config.predict_indices(fec,
tau_fraction=tau,
threshold=threshold)
# make sure the two methods are consistent
assert np.allclose(event_indices_1,event_indices_2) , \
"Programming error; FEATHER methods should be identical"
# POST: they are consistent. go ahead and plot force vs time, add lines
# where an event is predicted
print("Found events at indices: {}".format(event_indices_1))
# # Next, we show how to get do event detection based only on the approach
# We do the following steps(1-4), below
# (1) Create a new FEC based only on the retract
# (2) Save the FEC out to a csv
# (3) Run FEATHER on the CSV, just as we did above
_test_FEATHER_retract_only(fec, threshold, tau)
def run():
"""
<Description>
Args:
param1: This is the first param.
Returns:
This is a description of what is returned.
"""
# there are a couple of ways to call FEATHER in python.
# # (1) through an intermediate '.csv' file
in_file = '../Data/example_0.csv'
data = np.loadtxt(in_file, delimiter=',', skiprows=0)
header_info = _command_line_config._parse_csv_header(in_file)
time, sep, force = data[:, 0], data[:, 1], data[:, 2]
threshold = 1e-3
tau = 2e-2
meta_dict = dict(threshold=threshold,
tau=tau,
spring_constant=header_info.spring_constant,
trigger_time=header_info.trigger_time,
dwell_time=header_info.dwell_time)
event_indices_1 = _command_line_config.run_feather(in_file=in_file,
**meta_dict)
# # (2) directly, using python arrays and a constructed fec object. This is
# # likely to be much faster, since there is no extra file IO.
fec = _command_line_config.make_fec(time=time,
separation=sep,
force=force,
**meta_dict)
event_indices_2 = _command_line_config.predict_indices(fec,
tau_fraction=tau,
threshold=threshold)
# make sure the two methods are consistent
assert np.allclose(event_indices_1,event_indices_2) , \
"Programming error; FEATHEr methods are identical"
# POST: they are consistent. go ahead and plot force vs time, add lines
# where an event is predicted
print("Found events at indices: {}".format(event_indices_1))
plt.plot(time, force * (-1e12))
for i in event_indices_1:
plt.axvline(time[i])
plt.xlabel("Time (s)")
plt.ylabel("Force (pN)")
plt.show()
def parse_and_run():
"""
uses argparse to get the arguments _command_line_config.run_feather needs
Args:
none, but assumes called from command line properly
Returns:
event_indices, as predicted by FEATHER
"""
description = 'Predict event locations in a data file'
parser = argparse.ArgumentParser(description=description)
common = dict(required=True)
# # feathers options
parser.add_argument('-tau',
metavar='tau',
type=float,
help='tau fraction of curve (0,1)',
required=False,
default=1e-2)
parser.add_argument('-threshold',
metavar='threshold',
type=float,
help='probability threshold (0,1)',
**common)
# # 'meta' variables
parser.add_argument('-spring_constant',
metavar='spring_constant',
type=float,
help='spring constant of the probe',
**common)
parser.add_argument('-trigger_time',
metavar='trigger_time',
type=float,
help='time at which approach ends',
**common)
parser.add_argument('-dwell_time',
metavar='dwell_time',
type=float,
help='time between end of approach and retract start',
**common)
# path to the file
parser.add_argument('-file_input',
metavar="file_input",
type=str,
help="path to the force-extension curve file",
**common)
parser.add_argument('-file_output',
metavar="file_output",
type=str,
help="path to output the associated data",
**common)
# parse all the inputs
args = parser.parse_args()
out_file = os.path.normpath(args.file_output)
in_file = os.path.normpath(args.file_input)
# get the indices
feather_dict = dict(in_file=in_file,
threshold=args.threshold,
tau=args.tau,
spring_constant=args.spring_constant,
dwell_time=args.dwell_time,
trigger_time=args.trigger_time)
# call feather
event_indices = _command_line_config.run_feather(**feather_dict)
# done with the log file...
np.savetxt(fname=out_file,
delimiter=",",
newline="\n",
fmt="%d",
header="(C) PRH 2017\nEvent Indices",
X=event_indices)
return event_indices