"""eFEL Python API functions.
This module provides the user-facing Python API of eFEL.
The convenience functions defined here call the underlying 'cppcore' library
to hide the lower level API from the user.
Copyright (c) 2015, EPFL/Blue Brain Project
This file is part of eFEL <https://github.com/BlueBrain/eFEL>
This library is free software; you can redistribute it and/or modify it under
the terms of the GNU Lesser General Public License version 3.0 as published
by the Free Software Foundation.
This library is distributed in the hope that it will be useful, but WITHOUT
ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE. See the GNU Lesser General Public License for more
details.
You should have received a copy of the GNU Lesser General Public License
along with this library; if not, write to the Free Software Foundation, Inc.,
51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA.
"""
# pylint: disable=W0602,W0603,W0702, F0401, W0612, R0912
from __future__ import annotations
from pathlib import Path
from typing import Callable, Iterator, Literal, overload
from typing_extensions import deprecated
import numpy as np
import efel
import efel.cppcore as cppcore
import efel.pyfeatures as pyfeatures
from efel.pyfeatures.pyfeatures import get_cpp_feature
_settings = efel.Settings()
_int_settings = {}
_double_settings = {}
_string_settings = {}
[docs]
def reset():
"""Resets the efel to its initial state
The user can set certain values in the efel, like the spike threshold.
These values are persisten. This function will reset these value to their
original state.
"""
global _settings, _int_settings, _double_settings, _string_settings
_settings = efel.Settings()
_int_settings = {}
_double_settings = {}
_string_settings = {}
set_double_setting('spike_skipf', 0.1)
set_int_setting('max_spike_skip', 2)
set_double_setting('Threshold', _settings.threshold)
set_double_setting('DerivativeThreshold', _settings.derivative_threshold)
set_double_setting(
'DownDerivativeThreshold',
_settings.down_derivative_threshold)
set_double_setting('interp_step', 0.1)
set_double_setting('burst_factor', 1.5)
set_double_setting('strict_burst_factor', 2.0)
set_double_setting('voltage_base_start_perc', 0.9)
set_double_setting('voltage_base_end_perc', 1.0)
set_double_setting('current_base_start_perc', 0.9)
set_double_setting('current_base_end_perc', 1.0)
set_double_setting('rise_start_perc', 0.0)
set_double_setting('rise_end_perc', 1.0)
set_double_setting("initial_perc", 0.1)
set_double_setting("min_spike_height", 20.0)
set_int_setting("strict_stiminterval", 0)
set_double_setting("initburst_freq_threshold", 50)
set_double_setting("initburst_sahp_start", 5)
set_double_setting("initburst_sahp_end", 100)
set_int_setting("DerivativeWindow", 3)
set_str_setting("voltage_base_mode", "mean")
set_str_setting("current_base_mode", "mean")
set_double_setting("precision_threshold", 1e-10)
set_double_setting("sahp_start", 5.0)
set_int_setting("ignore_first_ISI", 1)
set_double_setting("impedance_max_freq", 50.0)
_initialise()
[docs]
def set_dependency_file_location(location: str | Path) -> None:
"""Sets the location of the Dependency file.
eFEL uses 'Dependency' files to let the user define versions of features to use.
The installation directory of eFEL contains a default 'DependencyV5.txt' file.
Unless users want to change this file, it is not necessary to call this function.
Modifying the Dependency file can be useful in debugging.
Args:
location: Path to the location of a Dependency file.
Raises:
FileNotFoundError: If the path to the dependency file doesn't exist.
"""
location = Path(location)
if not location.exists():
raise FileNotFoundError(f"Path to dependency file {location} doesn't exist")
_settings.dependencyfile_path = str(location)
[docs]
def get_dependency_file_location() -> str:
"""Gets the location of the Dependency file.
Returns:
Path to the location of a Dependency file.
"""
return _settings.dependencyfile_path
[docs]
def set_threshold(new_threshold: float) -> None:
"""Set the spike detection threshold in the eFEL, default -20.0
Args:
new_threshold: The new spike detection threshold value (in the same units
as the traces, e.g. mV).
"""
_settings.threshold = new_threshold
set_double_setting('Threshold', _settings.threshold)
[docs]
def set_derivative_threshold(new_derivative_threshold: float) -> None:
"""Set the threshold for the derivative for detecting the spike onset.
Some features use a threshold on dV/dt to calculate the beginning of an
action potential. This function allows you to set this threshold.
Args:
new_derivative_threshold: The new derivative threshold value (in the same units
as the traces, e.g. mV/ms).
"""
_settings.derivative_threshold = new_derivative_threshold
set_double_setting('DerivativeThreshold', _settings.derivative_threshold)
[docs]
def get_feature_names() -> list[str]:
"""Return a list with the name of all the available features
Returns:
A list that contains all the feature names available in
the eFEL. These names can be used in the feature_names
argument of e.g. get_feature_values()
"""
cppcore.Initialize(_settings.dependencyfile_path, "log")
feature_names: list[str] = []
cppcore.getFeatureNames(feature_names)
feature_names += pyfeatures.all_pyfeatures
return feature_names
[docs]
def feature_name_exists(feature_name: str) -> bool:
"""Returns True if the feature name exists in eFEL, False otherwise."""
return feature_name in get_feature_names()
def _get_feature(feature_name: str, raise_warnings=False) -> np.ndarray | None:
"""Get feature value, decide to use python or cpp"""
if feature_name in pyfeatures.all_pyfeatures:
return get_py_feature(feature_name)
else:
return get_cpp_feature(feature_name, raise_warnings=raise_warnings)
[docs]
def get_distance(
trace: dict,
feature_name: str,
mean: float,
std: float,
trace_check: bool = True,
error_dist: float = 250) -> float:
"""Calculate distance value for a list of traces.
Args:
trace: Trace dict that represents one trace. The dict should have the
following keys: 'T', 'V', 'stim_start', 'stim_end'
feature_name: Name of the the features for which to calculate the distance
mean: Mean to calculate the distance from
std: Std to scale the distance with
trace_check: Let the library check if there are spikes outside of stimulus
interval, default is True
error_dist: Distance returned when error, default is 250
Returns:
The absolute number of standard deviation the feature is away
from the mean. In case of anomalous results a value of
'error_dist' standard deviations is returned.
This can happen if: a feature generates an error, there are
spikes outside of the stimulus interval, the feature returns
a NaN, etc.
"""
_initialise()
# Next set time, voltage and the stimulus start and end
for item in list(trace.keys()):
cppcore.setFeatureDouble(item, [x for x in trace[item]])
if trace_check:
trace_check_success = get_feature_values(
[trace], ['trace_check'])[0] # type: ignore
if trace_check_success["trace_check"] is None:
return error_dist
feature_values = _get_feature(feature_name)
distance = 0.0
if feature_values is None or len(feature_values) < 1:
return error_dist
else:
# Am not using anything more fancy to avoid breaking exact
# reproducibility of legacy C++ code
for feature_value in feature_values:
distance += abs(feature_value - mean)
distance = distance / std / len(feature_values)
# Check for NaN
if distance != distance:
return error_dist
return distance
def _initialise() -> None:
"""Set cppcore initial values."""
cppcore.Initialize(_settings.dependencyfile_path, "log")
# flush the GErrorString from previous runs by calling getgError()
cppcore.getgError()
# First set some settings that are used by the feature extraction
for setting_name, int_setting in list(_int_settings.items()):
cppcore.setFeatureInt(setting_name, [int_setting])
for setting_name, double_setting in list(_double_settings.items()):
if isinstance(double_setting, list):
cppcore.setFeatureDouble(setting_name, double_setting)
else:
cppcore.setFeatureDouble(setting_name, [double_setting])
for setting_name, str_setting in list(_string_settings.items()):
cppcore.setFeatureString(setting_name, str_setting)
[docs]
def set_int_setting(setting_name: str, new_value: int) -> None:
"""Set a certain integer setting to a new value"""
_int_settings[setting_name] = new_value
[docs]
def set_double_setting(setting_name: str, new_value: float) -> None:
"""Set a certain double setting to a new value"""
_double_settings[setting_name] = new_value
[docs]
def set_str_setting(setting_name: str, new_value: str) -> None:
"""Set a certain string setting to a new value"""
_string_settings[setting_name] = new_value
@overload
def get_feature_values(
traces: list[dict],
feature_names: list[str],
parallel_map: Callable | None,
return_list: Literal[True],
raise_warnings: bool = True,
) -> list:
...
@overload
def get_feature_values(
traces: list[dict],
feature_names: list[str],
parallel_map: Callable | None,
return_list: Literal[False],
raise_warnings: bool = True,
) -> Iterator:
...
[docs]
def get_feature_values(
traces: list[dict],
feature_names: list[str],
parallel_map: Callable | None = None,
return_list: bool = True,
raise_warnings: bool = True,
) -> list | Iterator:
"""Calculate feature values for a list of traces.
This function is the core of eFEL API. A list of traces (in the form
of dictionaries) is passed as argument, together with a list of feature
names.
The return value consists of a list of dictionaries, one for each input
trace. The keys in the dictionaries are the names of the calculated
features, the corresponding values are lists with the feature values.
Beware that every feature returns an array of values. E.g. AP_amplitude
will return a list with the amplitude of every action potential.
Args:
traces: Every trace dict represents one trace. The dict should have the
following keys: 'T', 'V', 'stim_start', 'stim_end'
feature_names: List with the names of the features to be calculated on all
the traces.
parallel_map: Map function to parallelise over the traces. Default is the
serial map() function
return_list: By default the function returns a list of dicts. This
optional argument can disable this, so that the result of the
parallel_map() is returned. Can be useful for performance
reasons when an iterator is preferred.
raise_warnings: Raise warning when efel c++ returns an error
Returns:
For every input trace a feature value dict is returned (in
the same order). The dict contains the keys of
'feature_names', every key contains a numpy array with
the feature values returned by the C++ efel code.
The value is None if an error occured during the
calculation of the feature.
"""
if parallel_map is None:
parallel_map = map
traces_featurenames = ((trace, feature_names, raise_warnings) for trace in traces)
map_result = parallel_map(_get_feature_values_serial, traces_featurenames)
if return_list:
return list(map_result)
else:
return map_result
[docs]
def get_py_feature(feature_name: str) -> np.ndarray | None:
"""Return values of the given feature name."""
return getattr(pyfeatures, feature_name)()
def _get_feature_values_serial(
trace_featurenames: tuple[dict, list[str], bool]
) -> dict:
"""Single process of get_feature_values."""
trace, feature_names, raise_warnings = trace_featurenames
result = {}
if 'stim_start' in trace and 'stim_end' in trace:
try:
len(trace['stim_start'])
len(trace['stim_end'])
except BaseException:
raise Exception('Unable to determine length of stim_start or '
'stim_end, are you sure these are lists ?')
if len(trace['stim_start']) == 1 and len(trace['stim_end']) == 1:
if trace['stim_end'][0] <= trace['stim_start'][0]:
raise Exception(
'stim_end needs to be larger than '
'stim_start:\nstim_start=%f stim_end=%f' %
(trace['stim_start'][0], trace['stim_end'][0]))
else:
raise Exception(
'stim_start and stim_end in the trace '
'dictionary need to be lists of exactly 1 element')
else:
raise Exception('stim_start or stim_end missing from trace')
_initialise()
# Next set time, voltage and the stimulus start and end
for item in list(trace.keys()):
cppcore.setFeatureDouble(item, [x for x in trace[item]])
for feature_name in feature_names:
result[feature_name] = _get_feature(
feature_name, raise_warnings=raise_warnings)
return result
[docs]
def get_mean_feature_values(
traces: list[dict],
feature_names: list[str],
raise_warnings: bool = True) -> list[dict]:
"""Convenience function that returns mean values from get_feature_values()
Instead of return a list of values for every feature as get_feature_values()
does, this function returns per trace one value for every feature, namely
the mean value.
Args:
traces: Every trace dict represents one trace. The dict should have the
following keys: 'T', 'V', 'stim_start', 'stim_end'
feature_names: List with the names of the features to be calculated on all
the traces.
raise_warnings: Raise warning when efel c++ returns an error
Returns:
For every input trace a feature value dict is returned (in
the same order). The dict contains the keys of
'feature_names', every key contains the mean of the array
that is returned by get_feature_values()
The value is None if an error occured during the
calculation of the feature, or if the feature value array
was empty.
"""
featureDicts = get_feature_values(
traces,
feature_names,
parallel_map=None,
return_list=True,
raise_warnings=raise_warnings)
for featureDict in featureDicts: # type: ignore
for (key, values) in list(featureDict.items()):
if values is None or len(values) == 0:
featureDict[key] = None
else:
featureDict[key] = np.mean(values)
return featureDicts # type: ignore
reset()
# Deprecated functions
@deprecated("Use set_threshold instead")
def setThreshold(newThreshold: float) -> None:
set_threshold(newThreshold)
@deprecated("Use set_derivative_threshold instead")
def setDerivativeThreshold(newDerivativeThreshold: float) -> None:
set_derivative_threshold(newDerivativeThreshold)
@deprecated("Use get_feature_names instead")
def getFeatureNames() -> list[str]:
return get_feature_names()
@deprecated("Use feature_name_exists instead")
def FeatureNameExists(feature_name: str) -> bool:
return feature_name_exists(feature_name)
@deprecated("Use get_distance instead")
def getDistance(
trace,
featureName,
mean,
std,
trace_check=True,
error_dist=250) -> float:
return get_distance(trace, featureName, mean, std, trace_check, error_dist)
@deprecated("Use set_int_setting instead")
def setIntSetting(setting_name: str, new_value: int) -> None:
set_int_setting(setting_name, new_value)
@deprecated("Use set_double_setting instead")
def setDoubleSetting(setting_name: str, new_value: float) -> None:
set_double_setting(setting_name, new_value)
@deprecated("Use set_str_setting instead")
def setStrSetting(setting_name: str, new_value: str) -> None:
set_str_setting(setting_name, new_value)
@deprecated("Use get_feature_values instead")
def getFeatureValues(
traces,
featureNames,
parallel_map=None,
return_list=True,
raise_warnings=True):
return get_feature_values(
traces, featureNames, parallel_map, return_list, raise_warnings)
@deprecated("Use get_mean_feature_values instead")
def getMeanFeatureValues(
traces,
featureNames,
raise_warnings=True):
return get_mean_feature_values(traces, featureNames, raise_warnings)
@deprecated("Use get_dependency_file_location instead")
def getDependencyFileLocation() -> str:
return get_dependency_file_location()
@deprecated("Use set_dependency_file_location instead")
def setDependencyFileLocation(location: str | Path) -> None:
return set_dependency_file_location(location)