def dynamic_function_from_file(obj, filename):
"""
Load a dynamic function from a file and attach it to the obj. (A Network
or Trajectory.)
The filename must be <function_name>.py
"""
f = file(filename, 'r')
function_body = f.read()
f.close()
basename = os.path.basename(filename)
func = os.path.splitext(basename)[0]
file_type = os.path.splitext(basename)[1][1:]
if isinstance(obj, Trajectory_mod.Trajectory):
setattr(obj, '%s_functionBody' % func, function_body)
# We try to get the attribute 'namespace' for the object.
Network_mod._exec_dynamic_func(obj, func, getattr(obj, 'namespace', {}))
elif isinstance(obj, Network_mod.Network):
if file_type == 'py':
obj._dynamic_funcs_python[func] = function_body
obj.exec_dynamic_functions(disable_c = True)
elif file_type == 'c':
obj.exec_dynamic_functions(curr_c_code = function_body)
def dynamic_function_from_file(obj, filename):
"""
Load a dynamic function from a file and attach it to the obj. (A Network
or Trajectory.)
The filename must be <function_name>.py
"""
f = file(filename, 'r')
function_body = f.read()
f.close()
basename = os.path.basename(filename)
func = os.path.splitext(basename)[0]
file_type = os.path.splitext(basename)[1][1:]
if isinstance(obj, Trajectory_mod.Trajectory):
setattr(obj, '%s_functionBody' % func, function_body)
# We try to get the attribute 'namespace' for the object.
Network_mod._exec_dynamic_func(obj, func,
getattr(obj, 'namespace', {}))
elif isinstance(obj, Network_mod.Network):
if file_type == 'py':
obj._dynamic_funcs_python[func] = function_body
obj.exec_dynamic_functions(disable_c=True)
elif file_type == 'c':
obj.exec_dynamic_functions(curr_c_code=function_body)
def appendSensFromODEINT(self, timepoints, odeint_array, holds_dt=False):
if getattr(self, '_assignment', None) is None:
Network_mod._exec_dynamic_func(self,
'_assignment',
self.namespace,
bind=False)
if getattr(self, '_sens_assignment', None) is None:
Network_mod._exec_dynamic_func(self,
'_sens_assignment',
self.namespace,
bind=False)
numAdded = odeint_array.shape[0]
addedValues = scipy.zeros((numAdded, len(self.key_column)),
scipy.float_)
self.values = scipy.concatenate((self.values, addedValues))
self.timepoints = scipy.concatenate((self.timepoints, timepoints))
nDv = len(self.dynamicVarKeys)
nOv = len(self.optimizableVarKeys)
# fill in trajectory
for ii, dvId in enumerate(self.dynamicVarKeys):
self.values[-numAdded:, self.key_column.get(dvId)] = \
odeint_array[:, ii]
# ... and sensitivities
for ii, dvId in enumerate(self.dynamicVarKeys):
for jj, ovId in enumerate(self.optimizableVarKeys):
self.values[-numAdded:,
self.key_column.get((dvId, ovId))]\
= odeint_array[:, ii + (jj+1)*nDv]
self._assignment(self.values, self.timepoints, -numAdded, None)
self._sens_assignment(self.values, self.timepoints, -numAdded, None)
if holds_dt:
# fill in the time derivative of the trajectory
for ii, dvId in enumerate(self.dynamicVarKeys):
self.values[-numAdded:, self.key_column.get((dvId,'time'))] = \
odeint_array[:, ii + nDv*(nOv+1)]
# ... and of the sensitivities
for ii, dvId in enumerate(self.dynamicVarKeys):
for jj, ovId in enumerate(self.optimizableVarKeys):
self.values[-numAdded:,
self.key_column.get((dvId, ovId,'time'))]\
= odeint_array[:, ii + (jj+1)*nDv + nDv*(nOv+1)]
def appendSensFromODEINT(self, timepoints, odeint_array, holds_dt = False):
if getattr(self, '_assignment', None) is None:
Network_mod._exec_dynamic_func(self, '_assignment',
self.namespace, bind=False)
if getattr(self, '_sens_assignment', None) is None:
Network_mod._exec_dynamic_func(self, '_sens_assignment',
self.namespace, bind=False)
numAdded = odeint_array.shape[0]
addedValues = scipy.zeros((numAdded, len(self.key_column)),
scipy.float_)
self.values = scipy.concatenate((self.values, addedValues))
self.timepoints = scipy.concatenate((self.timepoints, timepoints))
nDv = len(self.dynamicVarKeys)
nOv = len(self.optimizableVarKeys)
# fill in trajectory
for ii, dvId in enumerate(self.dynamicVarKeys):
self.values[-numAdded:, self.key_column.get(dvId)] = \
odeint_array[:, ii]
# ... and sensitivities
for ii, dvId in enumerate(self.dynamicVarKeys):
for jj, ovId in enumerate(self.optimizableVarKeys):
self.values[-numAdded:,
self.key_column.get((dvId, ovId))]\
= odeint_array[:, ii + (jj+1)*nDv]
self._assignment(self.values, self.timepoints, -numAdded, None)
self._sens_assignment(self.values, self.timepoints, -numAdded, None)
if holds_dt :
# fill in the time derivative of the trajectory
for ii, dvId in enumerate(self.dynamicVarKeys):
self.values[-numAdded:, self.key_column.get((dvId,'time'))] = \
odeint_array[:, ii + nDv*(nOv+1)]
# ... and of the sensitivities
for ii, dvId in enumerate(self.dynamicVarKeys):
for jj, ovId in enumerate(self.optimizableVarKeys):
self.values[-numAdded:,
self.key_column.get((dvId, ovId,'time'))]\
= odeint_array[:, ii + (jj+1)*nDv + nDv*(nOv+1)]
def appendFromODEINT(self, timepoints, odeint_array, holds_dt = False):
if getattr(self, '_assignment', None) is None:
Network_mod._exec_dynamic_func(self, '_assignment',
self.namespace, bind=False)
numAdded = odeint_array.shape[0]
addedValues = scipy.zeros((numAdded, len(self.key_column)),
scipy.float_)
self.values = scipy.concatenate((self.values, addedValues))
self.timepoints = scipy.concatenate((self.timepoints, timepoints))
for ii, id in enumerate(self.dynamicVarKeys):
self.values[-numAdded:, self.key_column.get(id)] =\
odeint_array[:, ii]
self._assignment(self.values, self.timepoints, -numAdded, None)
if holds_dt :
for ii, id in enumerate(self.dynamicVarKeys) :
self.values[-numAdded:, self.key_column.get((id,'time'))] = \
odeint_array[:,ii+len(self.dynamicVarKeys)]
def createNetworkParameter(p):
id, name = p.getId(), p.getName()
v = p.getValue()
isConstant = p.getConstant()
parameter = Network_mod.Parameter(id=id,
value=v,
is_constant=isConstant,
name=name,
typical_value=None,
is_optimizable=True)
# optimizable by default
return parameter
def fromSBMLString(sbmlStr, id=None, duplicate_rxn_params=False):
r = libsbml.SBMLReader()
d = r.readSBMLFromString(sbmlStr)
if d.getNumErrors():
message = 'libSBML reported errors in SBML file. Try running file '\
'through the online validator: '\
'http://www.sbml.org/Facilities/Validator . Specific errors '\
'noted are: '
errors = []
for ii in range(d.getNumErrors()):
pm = d.getError(ii)
errors.append(pm.getMessage())
print(message + '; '.join(errors))
m = d.getModel()
modelId = m.getId()
if (id is None) and (modelId == ''):
raise ValueError('Network id not specified in SBML or passed in.')
elif id is not None:
modelId = id
rn = Network_mod.Network(id=modelId, name=m.getName())
for f in m.getListOfFunctionDefinitions():
id, name = f.getId(), f.getName()
math = f.getMath()
variables = []
for ii in range(math.getNumChildren() - 1):
variables.append(formula_to_py(math.getChild(ii)))
math = formula_to_py(math.getRightChild())
rn.addFunctionDefinition(id, variables, math)
for c in m.getListOfCompartments():
id, name = c.getId(), c.getName()
size = c.getSize()
isConstant = c.getConstant()
rn.addCompartment(id=id, size=size, isConstant=isConstant, name=name)
for s in m.getListOfSpecies():
id, name = s.getId(), s.getName()
compartment = s.getCompartment()
if s.isSetInitialConcentration():
iC = s.getInitialConcentration()
elif s.isSetInitialAmount():
iC = s.getInitialAmount()
else:
iC = 1
isBC, isConstant = s.getBoundaryCondition(), s.getConstant()
xml_text = s.toSBML()
uniprot_ids = set([
entry[1:].split('"')[0] for entry in xml_text.split('uniprot')[1:]
])
rn.addSpecies(id=id,
compartment=compartment,
initialConcentration=iC,
isConstant=isConstant,
is_boundary_condition=isBC,
name=name,
uniprot_ids=uniprot_ids)
for p in m.getListOfParameters():
parameter = createNetworkParameter(p)
rn.addVariable(parameter)
for rxn in m.getListOfReactions():
id, name = rxn.getId(), rxn.getName()
kL = rxn.getKineticLaw()
kLFormula = kL.getFormula()
substitution_dict = {}
# Deal with parameters defined within reactions
for p in kL.getListOfParameters():
parameter = createNetworkParameter(p)
# If a parameter with this name already exists, **and it has a
# different value than this parameter** we rename this parameter
# instance by prefixing it with the rxn name so there isn't a
# clash.
if parameter.id in rn.variables.keys():
logger.warn('Parameter %s appears in two different reactions '
'in SBML file.' % parameter.id)
if parameter.value != rn.variables.get(parameter.id).value or\
duplicate_rxn_params:
oldId = parameter.id
parameter.id = id + '_' + parameter.id
substitution_dict[oldId] = parameter.id
logger.warn('It has different values in the two positions '
'so we are creating a new parameter %s.' %
(parameter.id))
else:
logger.warn('It has the same value in the two positions '
'so we are only defining one parameter %s. '
'This behavior can be changed with the option '
'duplicate_rxn_params = True' % (parameter.id))
if parameter.id not in rn.variables.keys():
rn.addVariable(parameter)
kLFormula = ExprManip.sub_for_vars(kLFormula, substitution_dict)
# Assemble the stoichiometry. SBML has the annoying trait that
# species can appear as both products and reactants and 'cancel out'
# For each species appearing in the reaction, we build up a string
# representing the stoichiometry. Then we'll simplify that string and
# see whether we ended up with a float value in the end.
stoichiometry = {}
reactant_stoichiometry = {}
product_stoichiometry = {}
for reactant in rxn.getListOfReactants():
species = reactant.getSpecies()
stoichiometry.setdefault(species, '0')
stoich = reactant.getStoichiometryMath()
stoich = stoichToString(reactant, stoich)
stoichiometry[species] += '-(%s)' % stoich
if species in reactant_stoichiometry:
reactant_stoichiometry[species].append(stoich)
else:
reactant_stoichiometry[species] = [stoich]
for product in rxn.getListOfProducts():
species = product.getSpecies()
stoichiometry.setdefault(species, '0')
stoich = product.getStoichiometryMath()
stoich = stoichToString(product, stoich)
stoichiometry[species] += '+(%s)' % stoich
if species in product_stoichiometry:
product_stoichiometry[species].append(stoich)
else:
product_stoichiometry[species] = [stoich]
for species, stoich in stoichiometry.items():
stoich = ExprManip.simplify_expr(stoich)
try:
# Try converting the string to a float.
stoich = float(stoich)
except ValueError:
pass
stoichiometry[species] = stoich
for modifier in rxn.getListOfModifiers():
stoichiometry.setdefault(modifier.getSpecies(), 0)
rn.addReaction(id=id,
stoichiometry=stoichiometry,
kineticLaw=kLFormula,
reactant_stoichiometry=reactant_stoichiometry,
product_stoichiometry=product_stoichiometry)
for ii, r in enumerate(m.getListOfRules()):
if r.getTypeCode() == libsbml.SBML_ALGEBRAIC_RULE:
math = formula_to_py(r.getMath())
rn.add_algebraic_rule(math)
else:
variable = r.getVariable()
math = formula_to_py(r.getMath())
if r.getTypeCode() == libsbml.SBML_ASSIGNMENT_RULE:
rn.addAssignmentRule(variable, math)
elif r.getTypeCode() == libsbml.SBML_RATE_RULE:
rn.addRateRule(variable, math)
for ii, e in enumerate(m.getListOfEvents()):
id, name = e.getId(), e.getName()
if id == '':
id = 'event%i' % ii
try:
# For libSBML 3.0
trigger_math = e.getTrigger().getMath()
except AttributeError:
# For older versions
trigger_math = e.getTrigger()
trigger = formula_to_py(trigger_math)
if e.getDelay() is not None:
try:
# For libSBML 3.0
delay_math = e.getDelay().getMath()
except AttributeError:
# For older versions
delay_math = e.getDelay()
delay = formula_to_py(delay_math)
else:
delay = 0
timeUnits = e.getTimeUnits()
eaDict = KeyedList()
for ea in e.getListOfEventAssignments():
ea_formula = formula_to_py(ea.getMath())
ea_formula = ea_formula.replace('or(', 'or_func(')
ea_formula = ea_formula.replace('and(', 'and_func(')
eaDict.set(ea.getVariable(), ea_formula)
rn.addEvent(id=id,
trigger=trigger,
eventAssignments=eaDict,
delay=delay,
name=name)
for ii, con in enumerate(m.getListOfConstraints()):
id, name = con.getId(), con.getName()
if id == '':
id = 'constraint%i' % ii
trigger_math = con.getMath()
trigger = formula_to_py(trigger_math)
if con.isSetMessage():
message = con.getMessage()
else:
message = None
rn.addConstraint(id=id, trigger=trigger, message=message, name=name)
return rn
