UNAGI.train.runner.UNAGI_runner

class UNAGI.train.runner.UNAGI_runner(data_path, total_stage, iteration, trainer, idrem_dir, adversarial=True, GCN=True, connect_edges_cutoff=0.05)[source]

The UNAGI_runner class is used to set up the hyperparameters to run iDREM, find clustering optimal parameters and run the UNAGI model . It takes the following

Parameters:

  • data_path (the path to the data)

  • total_stage (the total number of time-series stages)

  • iteration (the total iteration to run the UNAGI model)

  • trainer (the trainer class to train the UNAGI model)

  • idrem_dir (the directory of the idrem software)

__init__(data_path, total_stage, iteration, trainer, idrem_dir, adversarial=True, GCN=True, connect_edges_cutoff=0.05)[source]

Methods

__init__(data_path, total_stage, iteration, ...)

annotate_stage_data(adata, stage, CPO)

Retreive the latent representations of given single cell data.

build_iteration_dataset()

Build the iteration dataset.

build_temporal_dynamics_graph()

Build the temporal dynamics graph.

load_stage_data()

Load the stage data from the data_path.

run(CPO)

Run the UNAGI pipeline.

run_CPO()

Find the clustering optimal parameters for the single cell data.

run_IDREM()

Run the iDREM software.

set_up_CPO(anchor_neighbors, max_neighbors, ...)

Set up the parameters for finding the clustering optimal parameters.

set_up_IDREM(...)

Set up the parameters for running the iDREM software.

set_up_species(species)

Set up the species for running the iDREM software.

update_cell_attributes(CPO)

Update and save the cell attributes including the top genes, cell types and latent representations.

update_gene_weights_table([topN])

Update the gene weights table.