def test_scipy(self):
global data
global data_mu
global data_sd
global init
global state_sd
global tspan
global sampler_args
all_data = {'OD600': pd.read_csv('data/OD600_Data.csv', header=[0, 1]),
'Fluor': pd.read_csv('data/GFP_Data.csv', header=[0, 1])
}
#Normalize Fluorescence by OD and convert to Molar
all_data['Fluor'] = divide(all_data['Fluor'], all_data['OD600'])
all_data['Fluor'] = multiply(all_data['Fluor'], 1e-6/(18.814*30))
data_by_model_num = {1: {'OD600': all_data['OD600'],
'Fluor': all_data['Fluor']
}
}
'''
Iterate across each type of sampler
'''
filename_samplers = [('settings_cf_de_1.ini', 'de', cf.scipy_differential_evolution),
('settings_cf_bh_1.ini', 'bh', cf.scipy_basinhopping),
('settings_cf_da_1.ini', 'da', cf.scipy_dual_annealing)
]
for filename, sampler, sampler_function in filename_samplers:
print('Test sampler', sampler)
#Extract the necessary information
#Note that subtraction of the blank has been included in this function!!!
data, data_mu, data_sd, init, state_sd, tspan = import_data(all_data, data_by_model_num)
sampler_args, config_data = sc.get_sampler_args(filename, sampler)
states = {1: ['OD600', 'Glu', 'Fluor'],
}
sampler_args = update_sampler_args(data, data_mu, data_sd, init, state_sd, tspan, sampler_args, states)
result = sampler_function(**sampler_args)
# assert type(result) == dict
print()
def test_get_sampler_args(self):
global user_core_models
global settings_file_1
global sampler_args
global config_data
filename = settings_file_1
#Get arguments for simulation
result = sc.get_sampler_args(filename, user_core_models=user_core_models)
#Expect models, params, config_data = result
assert len(result) == 2
try:
sampler_args, config_data = result
except:
sampler_args, config_data = result.values()
}
}
#Extract the necessary information
#Note that subtraction of the blank has been included in this function!!!
data, data_mu, data_sd, init, state_sd, tspan = import_data(
all_data, data_by_model_num)
'''
Iterate across each type of sampler
'''
filename_samplers = [('settings_de.ini', 'de',
cf.scipy_differential_evolution),
('settings_bh.ini', 'bh', cf.scipy_basinhopping),
('settings_da.ini', 'da', cf.scipy_dual_annealing)]
for filename, sampler, sampler_function in filename_samplers:
sampler_args, config_data = sc.get_sampler_args(filename, sampler)
states = {
1: ['OD600', 'Glu', 'Fluor'],
}
sampler_args = update_sampler_args(data, data_mu, data_sd, init,
state_sd, tspan, sampler_args,
states)
result = sampler_function(**sampler_args)
'''
The return value is a dictionary with the following values.
'a': The accepted samples as a DataFrame
'r': The rejected samples as a DataFrame if applicable
's': The scipy return value if applicable
directory = output_folder / 'simulation_results'
sim.export_simulation_results(ym, em, prefix=prefix, directory=directory)
'''Run curvefitting and model selection using the same configuration file.
The steps are as follows:
1. prepare configuration .ini file (aside from core model information)
- guess, priors, parameter_bounds, fixed_parameters,
2. get_sampler_args
3. update sampler_args with information from experimental data file
4. run curve fitting algorithm
5. plot results: can include simulation from guess (optional)
6. plot traces to check for convergence
7. calculate and rank aic
'''
sampler_args, config_data = sc.get_sampler_args(
model_files, user_core_models=user_core_models)
print('\nsampler_args:\n', sampler_args)
sampler_args['data'] = data_mu
for model_num in sampler_args['models']:
sampler_args['models'][model_num]['tspan'] = [tspan]
sampler_args['models'][model_num]['sd'] = state_sd
sampler_args['models'][model_num]['states'][-1] = 'Fluor/OD'
model_init = {
scenario: [
init[model_num][scenario].get(state, 0)
for state in sampler_args['models'][model_num]['states']
]
for scenario in init[model_num]
}
of analysis are all very similar. This allows you to copy and paste sections as
appropriate.
'''
'''
1. Reading the Settings File
'''
filename = 'settings_sa.ini'
config_data = sc.from_config(filename, sampler='sa')
'''
2. Compiling Arguments
'''
core_model = mh.from_config('Monod_Constitutive_Single.ini')
user_core_models = {core_model['system_type']: core_model}
sampler_args, config_data = sc.get_sampler_args(
config_data, sampler='sa', user_core_models=user_core_models)
'''
3. Wrapping for Models in Database
For models already in the database, we can combine the above steps into a single
function call.
'''
new_sampler_args, new_config_data = sc.get_sampler_args(filename,
sampler='sa')
'''
Note that possible values for sampler are:
'sa' : simulated annealing
'bh' : scipy's basin hopping
'da' : scipy's dual annealing
'de' : scipy's differential evoluation