def remove_temperature_constraints(ecmodel, etcmodel, prot):
# ecmodel: original model without temperature constraints
# etcmodel: model which have temperature constraints for some/all enzymes
# prot: uniprot id
# original coeffes
met1 = ecmodel.metabolites.get_by_id('prot_{0}'.format(prot))
old_kcat_coeffs = {
rxn.id: rxn.get_coefficient(met1)
for rxn in met1.reactions
}
met2 = ecmodel.metabolites.prot_pool
rxn = ecmodel.reactions.get_by_id('draw_prot_{0}'.format(prot))
old_fnt_coeff = rxn.get_coefficient(met2)
# rescue prot
met1 = etcmodel.metabolites.get_by_id('prot_{0}'.format(prot))
for rxn_id, old_coeff in old_kcat_coeffs.items():
rxn = etcmodel.reactions.get_by_id(rxn_id)
etc.change_rxn_coeff(rxn, met1, old_coeff)
met2 = etcmodel.metabolites.prot_pool
rxn = etcmodel.reactions.get_by_id('draw_prot_{0}'.format(prot))
etc.change_rxn_coeff(rxn, met2, old_fnt_coeff)
def do_a_fcc(model, tested_enzymes, enzymes):
# tsted_enzymes: a list of enzyme already test. In the order of test iteration
# 2. do fcc at each temperature
fccs = dict()
delta = 10
# Get all enzymes
try:
u0 = model.optimize().objective_value
except:
u0 = np.nan
print('Model Track u0:', T, u0)
for enz_id in enzymes:
if enz_id in tested_enzymes: continue
if np.isnan(u0) or u0 == 0: fccs[enz_id] = np.nan
else:
with model as m:
enz = m.metabolites.get_by_id(enz_id)
# Perturbe kcat in all reactions of this enzyme
for rxn in enz.reactions:
if rxn.id.startswith('draw_'): continue
new_coeff = rxn.get_coefficient(enz) / (1 + delta)
etc.change_rxn_coeff(rxn, enz, new_coeff)
try:
u = m.optimize().objective_value
except:
u = np.nan
if u is None: fc = np.nan
else: fc = (u - u0) / u0 / delta
fccs[enz.id] = fc
return fccs
def simulate_growth(model, Ts, sigma, df, prot):
'''
# model, cobra model
# Ts, a list of temperatures in K
# sigma, enzyme saturation factor
# df, a dataframe containing thermal parameters of enzymes: dHTH, dSTS, dCpu, Topt
# Ensure that Topt is in K. Other parameters are in standard units.
# prot: uniprotid, enzyme to be rescued
'''
rs0 = list()
rs = list()
met1 = model.metabolites.get_by_id('prot_{0}'.format(prot))
old_kcat_coeffs = {
rxn.id: rxn.get_coefficient(met1)
for rxn in met1.reactions
}
met2 = model.metabolites.prot_pool
rxn = model.reactions.get_by_id('draw_prot_{0}'.format(prot))
old_fnt_coeff = rxn.get_coefficient(met2)
for T in Ts:
with model:
# map temperature constraints
etc.map_fNT(model, T, df)
etc.map_kcatT(model, T, df)
etc.set_NGAMT(model, T)
etc.set_sigma(model, sigma)
try:
r = model.optimize().objective_value
except:
print('Failed to solve the problem')
r = 0
print(T - 273.15, r)
rs0.append(r)
# rescue prot
met1 = model.metabolites.get_by_id('prot_{0}'.format(prot))
for rxn_id, old_coeff in old_kcat_coeffs.items():
rxn = model.reactions.get_by_id(rxn_id)
etc.change_rxn_coeff(rxn, met1, old_coeff)
met2 = model.metabolites.prot_pool
rxn = model.reactions.get_by_id('draw_prot_{0}'.format(prot))
etc.change_rxn_coeff(rxn, met2, old_fnt_coeff)
try:
r = model.optimize().objective_value
except:
print('Failed to solve the problem')
r = 0
print(T - 273.15, r)
rs.append(r)
return rs0, rs
def do_fcc_at_T(T,params,sigma,delta):
# Ts: temperatures at which simulation will be performed
# params: a dataframe with Topt, Tm, T90, dCpt of all enzymes
# sigma: enzyme saturation factor
# delta: the fold change to be made for kcat of each enzyme
#
# return a dataframe with temperature as column and enzyme id as index
model = pickle.load(open('../models/aerobic.pkl','rb'))
# 1. Calculate thermal parameters
dfparam = etc.calculate_thermal_params(params)
enzymes = list(set([met.id for met in model.metabolites
if met.id.startswith('prot_') and met.id != 'prot_pool']))
# 2. do fcc at each temperature
data = dict()
etc.map_fNT(model,T,dfparam)
etc.map_kcatT(model,T,dfparam)
etc.set_NGAMT(model,T)
etc.set_sigma(model,sigma)
# Get all enzymes
try: u0 = model.optimize().objective_value
except: u0 = None
print('Model Track u0:',T,u0)
for enz_id in enzymes:
if u0 is None or u0 == 0: data[enz_id.split('_')[1]] = None
else:
with model as m:
enz = m.metabolites.get_by_id(enz_id)
# Perturbe kcat in all reactions of this enzyme
for rxn in enz.reactions:
if rxn.id.startswith('draw_'): continue
new_coeff = rxn.get_coefficient(enz)/(1+delta)
etc.change_rxn_coeff(rxn,enz,new_coeff)
try:u = m.optimize().objective_value
except: u = None
if u is None: fc = None
else: fc = (u-u0)/u0/delta
data[enz.id.split('_')[1]] = fc
# 3. Create a dataframe with temperature as column and enzyme ids as index
lst, index = [], []
for k,v in data.items():
lst.append(v)
index.append(k)
dfres = pd.DataFrame(data={T:lst},index=index)
return dfres
def simulate_one_particle(particle, T, ps):
growth_id = 'r_2111'
print('Load model')
mae = pickle.load(open('../models/aerobic.pkl', 'rb'))
# get ori coeffs for ERG1
met1 = mae.metabolites.get_by_id('prot_{0}'.format('P32476'))
old_kcat_coeffs = {
rxn.id: rxn.get_coefficient(met1)
for rxn in met1.reactions
}
met2 = mae.metabolites.prot_pool
rxn = mae.reactions.get_by_id('draw_prot_{0}'.format('P32476'))
old_fnt_coeff = rxn.get_coefficient(met2)
print('Set temperature constraits')
set_temperature_constraints(mae, particle, T)
erg_rxns = get_erg_rxns(mae)
print('Get vmax')
# This is to avoid some unbounded results
for rxn_id in erg_rxns:
mae.reactions.get_by_id(rxn_id).upper_bound = 1000
erg_vmax = {
rxn_id: get_maximal_flux_through_given_rxn(mae, rxn_id, growth_id)
for rxn_id in erg_rxns
}
rmax_vary_p = []
rmax_vary_p_rescue = []
print('Simulate down regulation')
for p in ps:
with mae:
down_regulate_erg_genes(mae, erg_vmax, p)
rmax_vary_p.append(mae.optimize().objective_value)
# rescue prot
for rxn_id, old_coeff in old_kcat_coeffs.items():
rxn = mae.reactions.get_by_id(rxn_id)
etc.change_rxn_coeff(rxn, met1, old_coeff)
rxn = mae.reactions.get_by_id('draw_prot_{0}'.format('P32476'))
etc.change_rxn_coeff(rxn, met2, old_fnt_coeff)
rmax_vary_p_rescue.append(mae.optimize().objective_value)
return rmax_vary_p, rmax_vary_p_rescue