def __init__(self, sbmlPath:str,
parametersToFit:typing.List[str]=None,
selectedColumns:typing.List[str]=None,
**kwargs):
"""
Parameters
----------
sbmlPath: path or URL to SBML files
selectedColumns: names of floating species used in fitting
parametersToFit: names of parameters to fit
kwargs: passed to ModelFitter constructor
"""
if "logger" in kwargs.keys():
self.logger = kwargs["logger"]
else:
self.logger = logs.Logger()
#
self.sbmlPath = sbmlPath
self.roadRunner = self._initializeRoadrunner()
self.selectedColumns = selectedColumns
self.parametersToFit = self._getSetableParameters(parametersToFit)
self.kwargs = kwargs
self.parametersToFit = self._getSetableParameters(parametersToFit)
self.parameterValueDct = {p: self.roadRunner[p]
for p in self.parametersToFit}
self.fitModelResult = TestHarnessResult()
self.bootstrapResult = TestHarnessResult()
self._validate()
def main(numIteration):
"""
Calculates the time to run iterations of the benchmark.
Parameters
----------
numIteration: int
Returns
-------
float: time in seconds
"""
logger = logs.Logger(logLevel=logs.LEVEL_STATUS, logPerformance=IS_TEST)
optimizerMethod = SBstoat.OptimizerMethod(SBstoat.METHOD_LEASTSQ,
{"max_nfev": 100})
fitter = mf.ModelFitter(
MODEL,
BENCHMARK_PATH,
["k1", "k2"],
selectedColumns=['S1', 'S3'],
isPlot=IS_PLOT,
logger=logger,
fitterMethods=[optimizerMethod],
bootstrapMethods=[optimizerMethod],
isProgressBar=False,
)
fitter.fitModel()
startTime = time.time()
fitter.bootstrap(numIteration=numIteration, isParallel=IS_PARALLEL)
elapsedTime = time.time() - startTime
if IS_TEST:
print(fitter.logger.formatPerformanceDF())
fitter.plotFitAll()
return elapsedTime
def runVirus(self, values):
ANTIMONY_MODEL = '''
// Equations
E1: T -> E ; beta*T*V ; // Target cells to exposed
E2: E -> I ; kappa*E ; // Exposed cells to infected
E3: -> V ; p*I ; // Virus production by infected cells
E4: V -> ; c*V ; // Virus clearance
E5: I -> ; delta*I // Death of infected cells
// Parameters - from the Influenza article,
beta = 3.2e-5; // rate of transition of target(T) to exposed(E) cells, in units of 1/[V] * 1/day
kappa = 4.0; // rate of transition from exposed(E) to infected(I) cells, in units of 1/day
delta = 5.2; // rate of death of infected cells(I), in units of 1/day
p = 4.6e-2; // rate virus(V) producion by infected cells(I), in units of [V]/day
c = 5.2; // rate of virus clearance, in units of 1/day
// Initial conditions
T = 4E+8 // estimate of the total number of susceptible epithelial cells
// in upper respiratory tract)
E = 0
I = 0
V = 0.75 // the dose of virus in TCID50 in Influenza experiment; could be V=0 and I = 20 instead for a natural infection
// Computed values
log10V := log10(V)
'''
dataSource = self.mkTimeSeries(values, "log10V")
parameterDct = dict(
beta=(0, 10e-5, 3.2e-5),
kappa=(0, 10, 4.0),
delta=(0, 10, 5.2),
p=(0, 1, 4.6e-2),
c=(0, 10, 5.2),
)
if IGNORE_TEST:
logger = logs.Logger(logLevel=logs.LEVEL_MAX)
else:
logger = LOGGER
study = ModelStudy(ANTIMONY_MODEL, [dataSource],
parameterDct=parameterDct,
selectedColumns=["log10V"],
doSerialize=False,
useSerialized=False,
logger=logger)
study.bootstrap(numIteration=100)
fitter = list(study.fitterDct.values())[0]
if IS_PLOT and (fitter.bootstrapResult is not None):
study.plotFitAll()
fitter = study.fitterDct["src_1"]
for name in parameterDct.keys():
if fitter.bootstrapResult is not None:
value = fitter.bootstrapResult.params.valuesdict()[name]
self.assertIsNotNone(value)
def main(maxNfev=MAX_NFEV):
"""
Calculates the time to run the benchmark.
Parameters
----------
numPopulation: int
Returns
-------
float: time in seconds
"""
logger = logs.Logger(logLevel=logs.LEVEL_SUPPRESS, logPerformance=IS_TEST)
models = [MODEL for _ in range(NUM_MODEL)]
parametersList = [PARAMETERS for _ in range(NUM_MODEL)]
optimizerMethod = SBstoat.OptimizerMethod(method="differential_evolution",
kwargs={
"popsize": 10,
'max_nfev': maxNfev
})
startTime = time.time()
suiteFitter = mkSuiteFitter(models,
OBSERVED_FILES,
parametersList,
MODEL_NAMES,
logger=logger,
fitterMethods=[optimizerMethod])
if NUM_FOLD == 1:
suiteFitter.fitSuite()
else:
suiteFitter.crossValidate(NUM_FOLD, isParallel=IS_PARALLEL)
elapsedTime = time.time() - startTime
if IS_TEST:
print(suiteFitter.reportFit())
if IS_PLOT:
suiteFitter.plotFitAll()
suiteFitter.plotResidualsSSQ()
return elapsedTime
def main(numIteration):
"""
Calculates the time to run iterations of the benchmark.
Parameters
----------
numIteration: int
Returns
-------
float: time in seconds
"""
logger = logs.Logger(logLevel=logs.LEVEL_MAX, logPerformance=IS_TEST)
fitter = ModelFitter(MODEL, BENCHMARK_PATH,
["k1", "k2"], selectedColumns=['S1', 'S3'], isPlot=IS_PLOT,
logger=logger)
fitter.fitModel()
startTime = time.time()
fitter.bootstrap(numIteration=numIteration, reportInterval=numIteration)
elapsedTime = time.time() - startTime
if IS_TEST:
print(fitter.logger.formatPerformanceDF())
fitter.plotFitAll()
return elapsedTime
import unittest
#matplotlib.use('TkAgg')
def remove(ffile):
if os.path.isfile(ffile):
os.remove(ffile)
IGNORE_TEST = False
IS_PLOT = False
TIMESERIES = th.getTimeseries()
DIR = os.path.dirname(os.path.abspath(__file__))
LOG_FILE = os.path.join(DIR, "testModelFitterBootstrap.log")
LOGGER = logs.Logger()
if IGNORE_TEST:
# Write log to std output
FITTER = th.getFitter(cls=mfb.ModelFitterBootstrap)
else:
FITTER = th.getFitter(cls=mfb.ModelFitterBootstrap, logger=LOGGER)
FITTER.fitModel()
NUM_ITERATION = 50
FILE_SERIALIZE = os.path.join(DIR, "modelFitterBootstrap.pcl")
FILES = [FILE_SERIALIZE]
MEAN_UNIFORM = 0.5 # Mean of uniform distribution
STD_UNIFORM = np.sqrt(1.0 / 12) # Standard deviation of uniform
remove(LOG_FILE) # Clean log file on each run
import unittest
#matplotlib.use('TkAgg')
def remove(ffile):
if os.path.isfile(ffile):
os.remove(ffile)
IGNORE_TEST = False
IS_PLOT = False
TIMESERIES = th.getTimeseries()
DIR = os.path.dirname(os.path.abspath(__file__))
LOG_FILE = os.path.join(DIR, "testModelFitterBootstrap.log")
LOGGER = logs.Logger(logLevel=logs.LEVEL_RESULT)
if IGNORE_TEST:
# Write log to std output
FITTER = th.getFitter(cls=mfb.ModelFitterBootstrap)
else:
FITTER = th.getFitter(cls=mfb.ModelFitterBootstrap, logger=LOGGER)
FITTER.fitModel()
NUM_ITERATION = 10
FILE_SERIALIZE = os.path.join(DIR, "modelFitterBootstrap.pcl")
FILES = [FILE_SERIALIZE]
MEAN_UNIFORM = 0.5 # Mean of uniform distribution
STD_UNIFORM = np.sqrt(1.0 / 12) # Standard deviation of uniform
remove(LOG_FILE) # Clean log file on each run
