sc_reconstruction.models.ReconPretrainedscGPT#

class sc_reconstruction.models.ReconPretrainedscGPT(checkpoint_path, gene_col='var_names', **kwargs)[source]#

Bases: object

Parameters:

  • checkpoint_path (str)

  • gene_col (str)

__init__(checkpoint_path, gene_col='var_names', **kwargs)[source]#

Pre-trained scGPT foundation model [Cui et al., 2024].

Wraps the scGPT encoder for ReconEval’s foundation-model reconstruction task. Requires the scgpt package and runs in the scgpt conda env.

Args:

checkpoint_path: Path to the pre-trained model checkpoint.

Parameters:

  • checkpoint_path (str)

  • gene_col (str)

Methods

__init__(checkpoint_path[, gene_col])

Pre-trained scGPT foundation model [Cui et al., 2024].

get_latent_representation(X)

prepare([adata])

Cache the adata reference and its gene list on the wrapper.

set_genes(genes)

Set the genes for reconstruction