corneto.methods.signaling.CellNOptDAG

class corneto.methods.signaling.CellNOptDAG(lambda_reg=0.001, max_flow=None, epsilon=1.0, backend=None)

Bases: FlowMethod

Infer a shared acyclic Boolean model from multiple conditions.

The method selects reactions globally and evaluates their Boolean truth in every condition. A single nonnegative flow has exactly the selected reaction dependencies as its internal support. Conservation, positive support, and acyclicity therefore require every selected dependency to lie on a path from a controlled input or inhibitor to a measured output.

Dummy vertices named AND<number> are compiled into one reaction per product. All operands of such a reaction share one selection variable and are evaluated conjunctively.

Parameters:

• lambda_reg (float) – Penalty for every selected reaction.

• max_flow (Optional[float]) – Upper bound for structural flow. By default, the number of compiled dependency edges is used.

• epsilon (float) – Minimum flow on every selected dependency edge.

• backend (Optional[Backend]) – Optimization backend.

__init__(lambda_reg=0.001, max_flow=None, epsilon=1.0, backend=None)

Parameters:

• lambda_reg (float)

• max_flow (float | None)

• epsilon (float)

• backend (Backend | None)

Methods

__init__([lambda_reg, max_flow, epsilon, ...])
build(pkn, *, inputs, measurements[, inhibitors])	Build a single-condition CellNOpt problem.
build_from_data(graph[, data])	Build the complete optimization problem.
build_many(pkn, *, inputs, measurements[, ...])	Build a problem for multiple named experimental conditions.
create_flow_based_problem(flow_problem, ...)	Add vectorized Boolean propagation and shared-flow selection.
create_problem(graph, data)	Create the optimization problem using a flow-based formulation.
description()	Return a short method description.
evaluate(*, inputs, measurements[, inhibitors])	Compare fixed-model predictions with supplied Boolean measurements.
fit(pkn, *, inputs, measurements[, ...])	Build and solve a shared Boolean model, returning self.
get_flow_bounds(graph, data)	Return bounds for the single shared structural flow.
get_selected_reaction_indices([threshold])	Return indices of reactions in the fixed solved model.
name()	Return the method name.
predict(*, inputs[, inhibitors])	Predict conditions from the fixed selected reactions.
preprocess(graph, data)	Compile reactions, validate condition data, and add flow boundaries.
references()	Returns citation keys for this method.
show_bibtex()	Display raw BibTeX entries in a formatted block for easy copying.
show_references()	Display formatted citations in a Jupyter notebook.

Attributes

backend

Return the optimization backend being used.

build(pkn, *, inputs, measurements, inhibitors=None)

Build a single-condition CellNOpt problem.

Parameters:

pkn (BaseGraph)

Return type:

ProblemDef

build_many(pkn, *, inputs, measurements, inhibitors=None)

Build a problem for multiple named experimental conditions.

Parameters:

pkn (BaseGraph)

Return type:

ProblemDef

fit(pkn, *, inputs, measurements, inhibitors=None, solve_options=None, **solver_options)

Build and solve a shared Boolean model, returning self.

Scientific arguments intentionally mirror build_many; backend options belong in solve_options (or can be supplied as solver keyword arguments). The solved problem remains available for adding constraints and inspecting expressions. Time-limited incumbents are accepted when they include finite reaction selections.

Parameters:

• pkn (BaseGraph)

• solve_options (Mapping[str, Any] | None)

• solver_options (Any)

Return type:

CellNOptDAG

preprocess(graph, data)

Compile reactions, validate condition data, and add flow boundaries.

Parameters:

• graph (BaseGraph)

• data (Data)

get_flow_bounds(graph, data)

Return bounds for the single shared structural flow.

Parameters:

• graph (BaseGraph)

• data (Data)

create_flow_based_problem(flow_problem, graph, data)

Add vectorized Boolean propagation and shared-flow selection.

Parameters:

• flow_problem (ProblemDef)

• graph (BaseGraph)

• data (Data)

Return type:

ProblemDef

get_selected_reaction_indices(threshold=0.5)

Return indices of reactions in the fixed solved model.

Parameters:

threshold (float)

Return type:

ndarray

predict(*, inputs, inhibitors=None)

Predict conditions from the fixed selected reactions.

Inputs and active inhibitors are clamped exactly as during training; every other species is the OR of its active selected reactions, whose literals are evaluated in a topological order. Held-out measurements are not accepted or consulted, and prediction never invokes a solver.

Return type:

Data

evaluate(*, inputs, measurements, inhibitors=None)

Compare fixed-model predictions with supplied Boolean measurements.

Return type:

dict[str, Any]

static name()

Return the method name.

Return type:

str

static description()

Return a short method description.

Return type:

str