bluepyemodel.emodel_pipeline.emodel_settings¶

EModelPipelineSettings class

Classes

EModelPipelineSettings([extraction_reader, ...])

Container for the settings used during the different steps of the e-model building pipeline.

class EModelPipelineSettings(extraction_reader=None, extraction_threshold_value_save=1, plot_extraction=True, pickle_cells_extraction=False, extract_absolute_amplitudes=False, rheobase_strategy_extraction='absolute', rheobase_settings_extraction=None, interpolate_RMP_extraction=False, default_std_value=0.001, bound_max_std=False, efel_settings=None, minimum_protocol_delay=0.0, stochasticity=False, morph_modifiers=None, threshold_based_evaluator=None, start_from_emodel=None, optimiser='IBEA', optimizer=None, optimisation_params=None, optimisation_timeout=600.0, threshold_efeature_std=None, max_ngen=100, optimisation_checkpoint_period=None, use_stagnation_criterion=True, validation_function='max_score', validation_threshold=5.0, use_ProbAMPANMDA_EMS=False, compile_mechanisms=False, n_model=3, optimisation_batch_size=5, max_n_batch=3, path_extract_config=None, name_Rin_protocol=None, name_rmp_protocol=None, validation_protocols=None, name_gene_map=None, plot_optimisation=True, plot_optimisation_progress=True, plot_parameter_evolution=True, plot_distributions=True, plot_scores=True, plot_traces=True, plot_thumbnail=True, plot_currentscape=True, plot_dendritic_ISI_CV=True, plot_dendritic_rheobase=True, plot_bAP_EPSP=False, plot_IV_curves=False, plot_FI_curve_comparison=False, plot_traces_comparison=False, run_plot_custom_sinspec=False, IV_curve_prot_name='iv', FI_curve_prot_name='idrest', plot_phase_plot=False, phase_plot_settings=None, sinespec_settings=None, currentscape_config=None, custom_bluepyefe_cells_pklpath=None, custom_bluepyefe_protocols_pklpath=None, save_recordings=False, neuron_dt=None, cvode_minstep=0.0, use_params_for_seed=True, current_precision=0.01, max_threshold_voltage=-30, strict_holding_bounds=True, max_depth_holding_search=7, max_depth_threshold_search=10, spikecount_timeout=50, files_for_extraction=None, targets=None, protocols_rheobase=None, auto_targets=None, auto_targets_presets=None)¶

Bases: object

Container for the settings used during the different steps of the e-model building pipeline. When using the “local” access point, these settings will be coming from the recipes.json file. When using the “nexus” access point, these settings will be coming from a resource of type EModelPipelineSettings.

This is a backend container class, not meant to be used directly by the user.

Creator of the EModelPipelineSettings class.

Parameters:

extraction_reader (function or list) – function used to read the ephys data during efeature extraction. If it is a list, it must contain the path to the file containing the function and name of the function. E.g: ["path_to_module", "name_of_function"]. If None, BluePyEfe will try to automatically use the correct default reader.
extraction_threshold_value_save (int) – during extraction, define the minimum number of values (data points) needed for an e-feature to be considered valid and be returned in the output of the extraction process. If the number of data point is lower than this value, the e-feature will be ignored.
plot_extraction (bool) – at the end of extraction, should the e-features and experimental traces be plotted.
pickle_cells_extraction (bool) – at the end of extraction, should the BluePyEfe objects be saved in pickle files for further analysis. The pickle files will be saved in the folder “./figures/{emodel_name}/efeatures_extraction/”.
extract_absolute_amplitudes (bool) – if True, during extraction, BluePyEfe will assume that the targets are expressed in absolute current amplitudes (nA) instead of the relative amplitudes when checking if a recording fits a given target.
rheobase_strategy_extraction (str) – during extraction, sets which function is used to compute the rheobase of the experimental cells. Can be ‘absolute’ (amplitude of the lowest current amplitude inducing at least a spike) or ‘flush’ (amplitude of the lowest current amplitude inducing at least a spike followed by another recording also inducing a spike).
rheobase_settings_extraction (dict) – settings related to the rheobase computation strategy. Keys have to match the arguments expected by the rheobase computation function present in the module bluepyefe.rheobase.
interpolate_RMP_extraction (bool) – whether to set the RMP after extraction as V_hold - R_in*I_Hold, which is an approximation of the RMP. This should be used when there is no protocol without holding current in the experimental data.
default_std_value (float) – At the end of e-features extraction, all features presenting a standard deviation of 0, will see their standard deviation replaced by the present value.
bound_max_std (bool) – If set to True, the std from extraction will be set to the mean value from extraction if it goes above it.
efel_settings (dict) – efel settings in the form {setting_name: setting_value} to be used during extraction. If settings are also informed in the targets on a per efeature basis, the latter will have priority.
minimum_protocol_delay (float) – during optimisation, if a protocol has an initial delay shorter than this value, the delay will be set to minimum_protocol_delay.
stochasticity (bool or list of str) – should channels behave stochastically if they can. If True, the mechanisms will be stochastic for all protocols. If a list of protocol names is provided, the mechanisms will be stochastic only for these protocols.
morph_modifiers (list of str or list of list) –
If str, name of the morph modifier to use from bluepyemodel.evaluation.modifiers. If List of morphology modifiers. Each modifier is defined by a list that includes the following elements: 1. The path to the file that contains the modifier. 2. The name of the function that applies the modifier. 3. Optionally, a “hoc_string” that represents the hoc code for the modifier. For example, morph_modifiers could be defined as follows:
morph_modifiers = [["path_to_module", "name_of_function", "hoc_string"], ...].
If the “hoc_string” is not provided, the system will search within the specified module for a string that matches the function name appended with “_hoc”. If None, the default modifier will replace the axon with a tappered axon initial segment (replace_axon_with_taper). If you do not wish to use any modifier, set the present argument to []. If ["bluepyopt_replace_axon"], the replace_axon function from bluepyopt.ephys.morphologies.NrnFileMorphology will be used and no other morph modifiers will be used.
threshold_based_evaluator (bool) – not used. To be deprecated.
start_from_emodel (dict) –
If informed, the optimisation for the present e-model will be instantiated using as values for the model parameters the ones from the e-model specified in the present dict. That option can be used for example to perform a two-steps optimisations. Example:
```
{
    "emodel": "bNAC",
    "etype": "bNAC",
    "iteration_tag": "mytest"
}.
```
optimiser (str) – name of the algorithm used for optimisation, can be “IBEA”, “SO-CMA” or “MO-CMA”.
optimizer (str) – here for backward compatibility. Use optimiser instead.

optimisation_params (dict) –

parameter for the optimisation process. Keys have to match the call of the optimiser. Here are the possible options and default values for the different optimisers:

- IBEA: {
    "offspring_size": 100 # number of individuals in the population
}
- SO-CMA: {
    "offspring_size": 20, # number of individuals in the population
    sigma=0.4 # initial standard deviation of the gaussian distribution
}
- MO-CMA: {
    "offspring_size": 20, # number of individuals in the population
    sigma=0.4, # initial standard deviation of the gaussian distribution
    weight_hv=0.5 # weight of the hypervolume score in the selection process
}

For more details, see the documentation of the bluepyopt.deapext package.

optimisation_timeout (float) – duration (in second) after which the evaluation of a protocol will be interrupted. When a protocol is interrupted, its response will be considered as invalid and the score of all the related e-features will be set to the maximum value (250 per default).
threshold_efeature_std (float) – if informed, the standard deviations of the efeatures will be thresholded at a minimum of abs(threshold_efeature_std * efeature_mean). Note that this will not overwrite the original standard deviations, but only modify them for the optimisation process.
max_ngen (int) – maximum number of generations of the evolutionary process of the optimisation.
optimisation_checkpoint_period (float) – minimum time (in s) between checkpoint save. None to save checkpoint independently of the time between them
use_stagnation_criterion (bool) – whether to use the stagnation stopping criterion on top of the maximum generation criterion during optimisation.
validation_function (str or list) – if str, can be max_score or mean_score. If list, must contain the path to the file containing the function and name of the function. E.g: ["path_to_module", "name_of_function"]
validation_threshold (float) – if max_score or mean_score were specified as the validation_function, this parameter will set the threshold under which the e-models will be considered to pass validation successfully.
optimisation_batch_size (int) – number of optimisation seeds to run in parallel. Only used by the Luigi workflow.
max_n_batch (int) – maximum number of optimisation batches. Only used by the Luigi workflow.
name_gene_map (str) – name of the gene mapping csv file. Only required when using the Nexus access_point and the IC-selector.
n_model (int) – minimum number of models to pass validation to consider the e-model building task done. Only used by the Luigi workflow.
use_ProbAMPANMDA_EMS (bool) – True to link ProbAMPANMDA_EMS in EMC on nexus, and download ProbAMPANMDA from nexus along with other mechanisms.
compile_mechanisms (bool) – should the mod files be copied in the local mechanisms_dir directory. Only used by the Luigi workflow.
path_extract_config (str) – specify the path to the .json file containing the targets for the extraction process. Defaults to the current directory. If an ‘iteration_tag’ is provided, the file will be located in ‘./run/iteration_tag’. This is only applicable with a local access point. See example emodel_pipeline_local_python for more details.
name_Rin_protocol (list or str) – name and amplitude of the protocol from which the input resistance should be selected from. The matching protocol should have “ohmic_input_resistance_vb_ssse” in its feature targets .E.g: ["IV", -20] or IV_-20. This setting has to be set before efeature extraction if you wish to run a threshold based evaluator.
name_rmp_protocol (list or str) – name and amplitude of the protocol from which the resting membrane potential should be selected from, e.g: ["IV", 0] or IV_0. The matching protocol should have “voltage_base” in its feature targets. This setting has to be set before efeature extraction if you wish to run a threshold based evaluator.
validation_protocols (list of str) – name of the protocols to be used for validation only. E.g. ["APWaveform_300"]. These protocols will not be used during optimisation.
plot_optimisation (bool) – legacy parameter, is not used anymore.
plot_optimisation_progress (bool) – True to plot the optimisation progress from checkpoint.
plot_parameter_evolution (bool) – during the plotting, should the evolution of the parameters be plotted.
plot_distributions (bool) – True to plot the parameters distributions.
plot_scores (bool) – True to plot the scores
plot_traces (bool) – True to plot the traces
plot_thumbnail (bool) – True to plot a trace that can be used as thumbnail
plot_currentscape (bool) – during the plotting, should the currentscapes be plotted for the recordings.
plot_dendritic_rheobase (bool) – True to plot dendritic rheobase (if present)
plot_bAP_EPSP (bool) – during the plotting, should ready-to-use back-propagating AP and EPSP protocols be run and plotted. Should be True only for pyramidal cells, since it depends on the presence of apical dendrite.
plot_IV_curves (bool) – during the plotting, should peak voltage and voltage_deflection IV curves be plotted for threshold-based sub-threshold IV protocols. Needs pickle_cells_extraction to be set to True.
plot_FI_curve_comparison (bool) – during the plotting, should FI curve be plotted for experimental and simulated data. Needs pickle_cells_extraction to be set to True.
plot_traces_comparison (bool) – True to plot a new figure with simulated traces on top of experimental traces. Needs pickle_cells_extraction to be set to True.
run_plot_custom_sinspec (bool) – True to run a SineSpec protocol, and plot
trace (its voltage and current)
impedance. (along with its)
IV_curve_prot_name (str) – which protocol to use to plot_IV_curves.
FI_curve_prot_name (str) – which protocol to use during plotting of FI curve comparison. The protocol should be supra-threshold and have the efeature mean_frequency associated to it.
plot_phase_plot (bool) – during the plotting, should phase plot be plotted.
phase_plot_settings (dict) –
settings for the phase plot. Should contain the following: “prot_names” (list of str): the names of the protocols to select for phase plot “amplitude” (float): amplitude of the protocol to select.

Only exactly this amplitude will be selected for model. An amplitude window is used for experimental trace selection

”amp_window” (float): amplitude window around amplitude for experimental
recording selection.Is not used for model trace selection

”relative_amp” (bool): Are amplitde and amp_window in relative amplitude (True)
or in absolute amplitude (False).

If set to None, it will default to: {

”prot_names”: [“idrest”], “amplitude”: 150, “amp_window”: 1.5, “relative_amp”: True,

}
sinespec_settings (dict) – contains amplitude settings for the SineSpec protocol, with keys ‘amp’ and ‘threshold_based’. ‘amp’ should be in percentage of threshold if ‘threshold_based’ is True, e.g. 150, or in nA if ‘threshold_based’ if false, e.g. 0.1. If set to None, it will default to: {“amp”: 0.05, “threshold_based”: False}
currentscape_config (dict) – currentscape configuration according to the currentscape documentation (https://github.com/BlueBrain/Currentscape). Note that current.names, output.savefig, output.fname and output.dir do not need to be set, since they are automatically overwritten by BluePyEModel. If current.names is set nonetheless, it will be used as the subset of available currents to be selected for the plot.
custom_bluepyefe_cells_pklpath (str) – file path to the cells.pkl output of BluePyEfe. If None, will use usual file path used in BluePyEfe, so this is to be set only to use a file at an unexpected path. Only used in plotting functions for IV curves, IF curves and phase plot.
custom_bluepyefe_protocols_pklpath (str) – file path to the protocols.pkl output of BluePyEfe. If None, will use usual file path used in BluePyEfe, so this is to be set only to use a file at an unexpected path. Only used in plotting functions for trace_comparison.
save_recordings (bool) – Whether to save the responses data under a folder named recordings.
neuron_dt (float) – time step of the NEURON simulator. If None, cvode will be used.
cvode_minstep (float) – minimum time step allowed when using cvode.
use_params_for_seed (bool) – use a hashed version of the parameter dictionary as a seed for the simulator
current_precision (float) – size of search interval in current to stop the search in SearchThresholdProtocol
max_threshold_voltage (float) – upper bound for the voltage during the search for the threshold or rheobase current (see SearchThresholdProtocol).
strict_holding_bounds (bool) – if True, the minimum and maximum values for the current used during the holding current search will be fixed. Otherwise, they will be widened dynamically if the holding current is beyond the initial bounds.
max_depth_holding_search (int) – maximum depth for the binary search for the holding current.
max_depth_threshold_search (int) – maximum depth for the binary search for the threshold current.
spikecount_timeout (float) – during the search of the threshold current, if the present timeout is reached, we set spikecount=2 as if many spikes were present, to speed up bisection search.
files_for_extraction (list) – temporary
targets (list) – temporary
protocols_rheobase (list) – temporary
auto_targets (list) – temporary
auto_targets_presets (list) – temporary