def _va_sim(model):
model.objective.direction = 'max'
sol_max = safe_optim(model)
model.objective.direction = 'min'
sol_min = safe_optim(model)
return sol_min, sol_max
def _va_sim(model, rxn, epsilon=EPSILON):
lb, ub = rxn.bounds
rxn.lower_bound = epsilon
# model.objective.direction = 'max'
sol_max = safe_optim(model)
rxn.lower_bound = lb
rxn.upper_bound = -epsilon
# model.objective.direction = 'min'
sol_min = safe_optim(model)
rxn.upper_bound = ub
return ~isnan(sol_min.objective_value), \
~isnan(sol_max.objective_value)
def simulate(available_uptake, model, variables, warm_start=None):
model.logger.info('available_uptake = {}'.format(available_uptake))
model.reactions.EX_glc__D_e.lower_bound = available_uptake
model.growth_reaction.lower_bound = 0
model.growth_reaction.upper_bound = 10
model.objective = model.growth_reaction.id
model.objective.direction = 'max'
out = safe_optim(model)
if model.solver.status == 'infeasible' or model.solver.status == 'time_limit':
ret = {
'obj': np.nan,
'mu': np.nan,
'mu_lb': np.nan,
'mu_ub': np.nan,
'available_substrate': available_uptake,
'uptake': np.nan,
'prot_ratio': np.nan,
'mrna_ratio': np.nan
}
for var in variables:
ret[var + '_lb'] = np.nan
ret[var + '_ub'] = np.nan
print('INFEASIBLE SOLUTION AT q={}'.format(available_uptake))
return pd.Series(ret)
growth_solution = copy(model.solution)
mu_i, mu_lb, mu_ub = get_active_growth_bounds(model)
mu = model.growth_reaction.flux
# release_warm_start(model)
try:
prot_ratio = model.interpolation_variable.prot_ggdw.variable.primal
mrna_ratio = model.interpolation_variable.mrna_ggdw.variable.primal
except AttributeError:
# Model without Neidhardt data
prot_ratio = np.nan
mrna_ratio = np.nan
ret = {
'obj': model.solution.objective_value,
'mu': mu,
'mu_lb': mu_lb,
'mu_ub': mu_ub,
'available_substrate': -1 * available_uptake,
'uptake': -1 * growth_solution.fluxes['EX_glc__D_e'],
'prot_ratio': prot_ratio,
'mrna_ratio': mrna_ratio
}
fix_growth(model, model.solution)
for var in variables:
model.objective = model.variables.get(var)
lb, ub = _va_sim(model)
ret[var + '_lb'] = lb.objective_value
ret[var + '_ub'] = ub.objective_value
print(pd.Series(ret))
release_growth(model)
# apply_warm_start(model, growth_solution)
return pd.Series(ret)