dyngpt package

Subpackages

• dyngpt.dynmodels package
  • Submodules
  • dyngpt.dynmodels.afl module
    • get_afl_model_config()
  • dyngpt.dynmodels.arl module
    • get_arl_model_config()
  • dyngpt.dynmodels.default_model module
    • get_default_model_config()
  • dyngpt.dynmodels.isc module
    • get_isc_model_config()
  • dyngpt.dynmodels.nm_nm module
    • get_nm_nm_model_config()
  • dyngpt.dynmodels.on_off_nm module
    • get_on_off_nm_model_config()
  • dyngpt.dynmodels.sdp module
    • get_sdp_model_config()
  • dyngpt.dynmodels.sir module
    • get_sir_model_config()
  • dyngpt.dynmodels.sirs module
    • get_sirs_model_config()
  • dyngpt.dynmodels.toggle_switch module
    • get_toggle_switch_model_config()
  • dyngpt.dynmodels.ts_mp module
    • get_ts_mp_model_config()
  • dyngpt.dynmodels.ts_txl module
    • get_ts_txl_model_config()
  • Module contents
• dyngpt.plotting package
  • Module contents
• dyngpt.tools package
  • Module contents

Submodules

dyngpt.inferrer module

dyngpt.inferrer.inferring_dynamics(observed_data, config_infer, gene_names=[], true_params=[], synthetic_flag=0, new_model=False, synthetic_data=[])

Infer the underlying dynamic parameters of a state transition network from observed data.

Args:

observed_data (numpy.ndarray): The observed dataset containing state counts. config_infer (Namespace): Configuration object containing inference parameters. gene_names (list, optional): List of gene names corresponding to the observed data. Defaults to []. true_params (list, optional): List of true parameter values for validation. Defaults to []. synthetic_flag (int, optional): Flag indicating whether synthetic data is used (1) or not (0). Defaults to 0. new_model (bool, optional): Whether to initialize a new model or use a pre-trained model. Defaults to False. synthetic_data (list, optional): Synthetic dataset for model validation, if applicable. Defaults to [].

Returns:

dict: Contains the inferred parameters, loss values, KL divergence values, and other evaluation metrics.

dyngpt.solver module

dyngpt.solver.fine_tune_training(config_fine_tune)

Fine-tune the neural network model using a specified training configuration.

Args:

config_fine_tune (Namespace): Configuration object containing hyperparameters,

dataset paths, training settings, and model specifications.

Raises:

ValueError: If an unknown loss type is specified in args.loss_type.

Returns:

dict: A dictionary containing training results, including loss means and the path

to the fine-tuned model weights.

dyngpt.solver.pre_training(config_pretrain)

Pre-train the neural network model for state transition network.

Args:

config_pretrain (Namespace): Configuration object containing training parameters.

Returns:

dict: A dictionary containing training loss means and the final model weight path.

dyngpt.solver.solving_STN(model_name, params=[], initial_value=[], args='')

Solve the state transition network using a trained neural network model.

Args:

model_name (str): The name of the model to be used. params (list, optional): A list of parameters for the model. Defaults to []. initial_value (list, optional): A list of initial values for the system. Defaults to []. args (Namespace, optional): Additional arguments containing model configurations,

including device, batch size, and initial values. Defaults to “”.

Returns:

np.ndarray: Sampled data generated by the neural network based on the provided parameters.

dyngpt.solver.solving_state_transition_network(model_name, params=[], initial_value=[], args='')

Module contents