cascade.sim.simulate_regimes

cascade.sim.simulate_regimes(dag, design, interv, random_state=None)[source]

Simulate interventional data based on a causal structure with multiple sets of intervention effect in parallel

Parameters:

  • dag (DiGraph) – A directed acyclic graph representing the causal structure

  • design (Mapping[Targets, int]) – A mapping from intervention targets to sample numbers

  • interv (Mapping[str, float | Callable[[int, RandomState | int | None], Sequence[float]]]) – Intervention scaling factor \(\lambda\) of each intervention target or sampler function of such (\(\lambda = 0\) for knockout, \(0 \lt \lambda \lt 1\) for knockdown, \(\lambda \gt 1\) for knockup)

  • random_state (RandomState | int | None) – Random state

Return type:

AnnData

Returns:

Simulated dataset

Note

  • The signal-to-noise ratio for each simulated variable should be provided as a node attribute in dag called "snr".

  • The activation function for each simulated variable should be provided as a node attribute in dag called "act".