def test_evaluate_performance_with_combined_protocol(self):
config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=protocols.SingleActionPotentialProtocol(),
tunable_parameters=[
ga_configs.Parameter(name='g_b_na', default_value=0.95),
ga_configs.Parameter(name='g_na', default_value=3671.2302)
],
params_lower_bound=0.9,
params_upper_bound=1.1,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.15,
tournament_size=2)
single_ap_ga = parameter_tuning_genetic_algorithm.\
ParameterTuningGeneticAlgorithm(config=config)
single_ap_error = single_ap_ga._evaluate_performance(
new_parameters=[0.94, 3000.])
config.secondary_protocol = protocols.IrregularPacingProtocol(
duration=5, stimulation_offsets=[0.1, 0.2])
combined_protocol_ga = parameter_tuning_genetic_algorithm.\
ParameterTuningGeneticAlgorithm(config=config)
combined_protocol_error = combined_protocol_ga._evaluate_performance(
new_parameters=[0.94, 3000.])
# Combined protocol error should be larger than a single protocol error.
self.assertGreater(combined_protocol_error, single_ap_error)
def setUp(self):
config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=protocols.SingleActionPotentialProtocol(),
tunable_parameters=[
ga_configs.Parameter(name='g_b_na', default_value=1.),
ga_configs.Parameter(name='g_na', default_value=3600.)
],
params_lower_bound=0.9,
params_upper_bound=1.1,
mate_probability=0.9,
mutate_probability=1.,
gene_swap_probability=0.5,
gene_mutation_probability=0.15,
tournament_size=2)
self.ga_result = genetic_algorithm_results.GAResultParameterTuning(
config=config)
generation_one = [
genetic_algorithm_results.ParameterTuningIndividual(
parameters=[0.95, 3671.2302], fitness=0),
genetic_algorithm_results.ParameterTuningIndividual(
parameters=[0.1, 3400.0], fitness=2),
]
generation_two = [
genetic_algorithm_results.ParameterTuningIndividual(
parameters=[0.6, 3500.0], fitness=5),
genetic_algorithm_results.ParameterTuningIndividual(
parameters=[0.9, 3600.0], fitness=9),
]
self.ga_result.generations.append(generation_one)
self.ga_result.generations.append(generation_two)
def test_run_comparing_experiment_raises_value_error_double_protocol(self):
parameters = [
ga_configs.Parameter(name='g_na', default_value=3671.2302),
ga_configs.Parameter(name='g_f_s', default_value=30.10312),
]
sap_protocol = protocols.SingleActionPotentialProtocol()
sap_config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=sap_protocol,
tunable_parameters=parameters,
params_lower_bound=0.5,
params_upper_bound=1.5,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.1,
tournament_size=2)
configs = generate_config_list()
configs.append(sap_config)
with self.assertRaises(ValueError) as ve:
parameter_tuning_experiments.run_comparison_experiment(
configs=configs, iterations=2)
self.assertEqual('Configs do not have unique protocols.',
str(ve.exception))
def test_make_parameter_scaling_examples(self):
params = [[1, 2], [1.5, 3]]
protocol_type = 'SAP'
default_params = [
ga_configs.Parameter(name='g_na', default_value=15),
ga_configs.Parameter(name='g_ca', default_value=1),
]
examples = parameter_tuning_experiments._make_parameter_scaling_examples(
params=params,
protocol_type=protocol_type,
default_params=default_params)
expected_examples = [[1, 'g_na', 'SAP'], [2, 'g_ca', 'SAP'],
[1.5, 'g_na', 'SAP'], [3, 'g_ca', 'SAP']]
self.assertListEqual(examples, expected_examples)
def setUp(self):
config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=protocols.SingleActionPotentialProtocol(),
tunable_parameters=[
ga_configs.Parameter(name='g_b_na', default_value=0.95),
ga_configs.Parameter(name='g_na', default_value=3671.2302)
],
params_lower_bound=0.9,
params_upper_bound=1.1,
mutate_probability=1.0,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.15,
tournament_size=2)
self.ga = parameter_tuning_genetic_algorithm.\
ParameterTuningGeneticAlgorithm(config=config)
def test_update_parameters(self):
protocol = protocols.SingleActionPotentialProtocol()
tunable_parameters = [
ga_configs.Parameter(name='G_Na', default_value=1),
ga_configs.Parameter(name='G_F', default_value=1),
ga_configs.Parameter(name='G_Ks', default_value=1),
ga_configs.Parameter(name='G_Kr', default_value=1),
ga_configs.Parameter(name='G_K1', default_value=1),
ga_configs.Parameter(name='G_bNa', default_value=1),
ga_configs.Parameter(name='G_NaL', default_value=1),
ga_configs.Parameter(name='G_CaL', default_value=1),
ga_configs.Parameter(name='G_pCa', default_value=1),
ga_configs.Parameter(name='G_bCa', default_value=1)
]
new_params = [.7, .7, .7, .7, .7, .7, .7, .7, .7, .7]
baseline_trace = paci_2018.generate_trace(protocol)
new_trace = paci_2018.generate_trace(protocol, tunable_parameters,
new_params)
tst.assert_raises(AssertionError, tst.assert_array_equal,
baseline_trace.y, new_trace.y,
'updating parameters does not change trace')
def generate_config_list():
parameters = [
ga_configs.Parameter(name='g_na', default_value=3671.2302),
ga_configs.Parameter(name='g_f_s', default_value=30.10312),
]
sap_protocol = protocols.SingleActionPotentialProtocol()
ip_protocol = protocols.IrregularPacingProtocol(duration=3,
stimulation_offsets=[0.1])
vc_protocol = protocols.VoltageClampProtocol(steps=[
protocols.VoltageClampStep(duration=0.1, voltage=-0.08),
protocols.VoltageClampStep(duration=0.1, voltage=-0.12),
protocols.VoltageClampStep(duration=0.5, voltage=-0.06),
protocols.VoltageClampStep(duration=0.05, voltage=-0.04),
protocols.VoltageClampStep(duration=0.15, voltage=0.02),
protocols.VoltageClampStep(duration=0.025, voltage=-0.08),
protocols.VoltageClampStep(duration=0.3, voltage=0.04)
])
sap_config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=sap_protocol,
tunable_parameters=parameters,
params_lower_bound=0.5,
params_upper_bound=1.5,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.1,
tournament_size=2)
ip_config = ga_configs.ParameterTuningConfig(population_size=2,
max_generations=2,
protocol=ip_protocol,
tunable_parameters=parameters,
params_lower_bound=0.5,
params_upper_bound=1.5,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.1,
tournament_size=2)
vc_config = ga_configs.ParameterTuningConfig(population_size=2,
max_generations=2,
protocol=vc_protocol,
tunable_parameters=parameters,
params_lower_bound=0.5,
params_upper_bound=1.5,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.1,
tournament_size=2)
combined_config = ga_configs.ParameterTuningConfig(
population_size=2,
max_generations=2,
protocol=ip_protocol,
tunable_parameters=parameters,
params_lower_bound=0.5,
params_upper_bound=1.5,
mutate_probability=1.,
mate_probability=0.9,
gene_swap_probability=0.5,
gene_mutation_probability=0.1,
tournament_size=2,
secondary_protocol=vc_protocol)
return [sap_config, ip_config, vc_config, combined_config]
"""Main driver for program. Before running, make sure to have a directory
called `figures` for matplotlib pictures to be stored in."""
import copy
import time
import parameter_tuning_experiments
import protocols
import ga_configs
import genetic_algorithm_results
import voltage_clamp_optimization_experiments
PARAMETERS = [
ga_configs.Parameter(name='G_Na', default_value=1),
ga_configs.Parameter(name='G_F', default_value=1),
ga_configs.Parameter(name='G_Ks', default_value=1),
ga_configs.Parameter(name='G_Kr', default_value=1),
ga_configs.Parameter(name='G_K1', default_value=1),
ga_configs.Parameter(name='G_bNa', default_value=1),
ga_configs.Parameter(name='G_NaL', default_value=1),
ga_configs.Parameter(name='G_CaL', default_value=1),
ga_configs.Parameter(name='G_pCa', default_value=1),
ga_configs.Parameter(name='G_bCa', default_value=1),
ga_configs.Parameter(name='K_NaCa', default_value=1)
]
# Parameters are sorted alphabetically to maintain order during each
# generation of the genetic algorithm.
PARAMETERS.sort(key=lambda x: x.name)
SAP_PROTOCOL_KERNIK = protocols.SingleActionPotentialProtocol(1800)