src.model_balancing.model_balancing_cvx

A module for performing convex model balancing (using CVXPY).

Module Contents

class src.model_balancing.model_balancing_cvx.ModelBalancingConvex(S: numpy.array, fluxes: numpy.array, A_act: numpy.array, A_inh: numpy.array, metabolite_names: List[str], reaction_names: List[str], state_names: List[str], rate_law: str = 'CM', solver: str = 'MOSEK', beta: float = 0.0, **kwargs)

Bases: object

A class for performing Convex Model Balancing (using CVXPY).

All input parameters are provided through the constructor. Then running .solve() will find the (global) optimum and the result will be stored in the cvxpy.Variable objects that represent the independent variables: .ln_Km .ln_Ka .ln_Ki .ln_Keq .ln_kcatf .ln_conc_met and also in the two dependent cvxpy.Expression objects: .ln_kcatr .ln_conc_enz

Use the .value to get the solution for any of these Variables/Expressions. Alternatively, .to_model_sbtab and .to_state_sbtab return SBtab objects containing the state (met_conc, enz_conc) and the model (Km, Ki, Ka, Keq, kcatf, kcatr) values respectively.

S

fluxes

A_act

A_inh

metabolite_names

reaction_names

state_names

rate_law = 'CM'

beta = 0.0

solver = 'MOSEK'

Ncond

ln_geom_mean

z2_scores

ln_lower_bound

ln_upper_bound

total_z2_scores

property driving_forces: cvxpy.Expression

Calculates the driving forces of all reactions.

property ln_kcatr: cvxpy.Expression

Calculate the kcat-reverse based on Haldane relationship constraint.

property ln_eta_thermodynamic: cvxpy.Expression

Calculate the thermodynamic term of the enzyme \(\eta^{thr}\).

property ln_eta_kinetic: cvxpy.Expression

Calculate the kinetic (saturation) term of the enzyme \(\eta^{kin}\).

property ln_eta_regulation: cvxpy.Expression

Calculate the regulation (allosteric) term of the enzyme \(\eta^{reg}\).

property ln_conc_enz: cvxpy.Expression

Calculate the required enzyme levels based on fluxes and rate laws.

penalty_term_beta(ln_Km: numpy.array | cvxpy.Expression, ln_conc_met: numpy.array | cvxpy.Expression) → cvxpy.Expression

Calculate the penalty term for c/Km.

is_gmean_feasible() → bool

Check if the gmean is a thermodynamically feasible solution.

This is useful because we sometimes would like to initialize the optimization with the geometric means, but that can only be done if that point is feasible (otherwise, the dependent parameter conc_enz is not defined).

initialize_with_gmeans() → None

Initialize the independent parameters with their gmeans.

Note that the dependent parameters (kcatr and ln_conc_enz) can both be very far from their gmeans, and that the system might not be thermodynamically feasible.

solve(verbose: bool = False) → str

Use CVXPY to find the global optimum that minimizes all z-scores.

find_inner_point(verbose: bool = False) → Any

Find a point insize the feasible thermodynamic space.

If the geometric mean is not a feasible solution, we will need this function in order to initialize the solver with a point inside the thermodynamically feasible space.

property objective_value: float

Get the objective value (i.e. the sum of squares of all z-scores).

get_z_scores() → Dict[str, numpy.array]

print_z_scores(precision: int = 2) → None

Print the z-score values for all variables.

print_status() → None

Print a status report based on the current solution.

to_state_sbtab() → sbtab.SBtab.SBtabDocument

Create a state SBtab.

The state SBtab contains the values of the state-dependent variables, i.e. flux, concentrations of metabolites, concentrations of enzymes, and the ΔG’ values.

to_model_sbtab() → sbtab.SBtab.SBtabDocument

Create a model SBtab.

The model SBtab contains the values of the state-independent variables, i.e. kcatf, kcatr, Km, Ka, and Ki.