descope.dataset

class BaseDataset(adata: str | AnnData, pert_col: str = 'perturbation', ctrl_name: str = 'control', perts_to_include: list | None = None, perts_to_exclude: list | None = None, gene_embs_file: str = './ESM2_pert_features.pt')

Bases: Dataset, ABC

MAIN_INPUT_NAME = None

RANDOM_MAPPING_CONTROL_TO_CONTROL = False

get_ctrl_cell_indices(adata: AnnData) → list[int]

abstract preprocess_adata(adata: AnnData) → AnnData

class DatasetForATAC(adata: str | AnnData, pert_col: str = 'perturbation', ctrl_name: str = 'control', topk_ccres: int = 50000, perts_to_include: list | None = None, perts_to_exclude: list | None = None, gene_embs_file: str = './ESM2_pert_features.pt')

Bases: BaseDataset

MAIN_INPUT_NAME = 'ctrl_cell_tf_idf'

RANDOM_MAPPING_CONTROL_TO_CONTROL = False

preprocess_adata(adata: AnnData) → AnnData

class DatasetForRNA(adata: str | AnnData, pert_col: str = 'target_gene', ctrl_name: str = 'non-targeting', target_sum: float = 10000.0, skip_raw_counts_check: bool = False, perts_to_include: list | None = None, perts_to_exclude: list | None = None, gene_embs_file: str = './ESM2_pert_features.pt')

Bases: BaseDataset

MAIN_INPUT_NAME = 'ctrl_cell_expr'

RANDOM_MAPPING_CONTROL_TO_CONTROL = False

preprocess_adata(adata: AnnData) → AnnData

class HFBaseDataset(hf_dataset: Dataset, ctrl_name: str = 'control', gene_embs_file: str = './ESM2_pert_features.pt', mse_weights_pkl_file: str | None = None)

Bases: Dataset

MAIN_INPUT_NAME = None

RANDOM_MAPPING_CONTROL_TO_CONTROL = False

static collate_fn(batch)

get_ctrl_cell_indices_for_each_celltype() → dict[str, list[int]]

class HFDatasetForATAC(hf_dataset: Dataset, ctrl_name: str = 'control', gene_embs_file: str = './ESM2_pert_features.pt', mse_weights_pkl_file: str | None = None)

Bases: HFBaseDataset

MAIN_INPUT_NAME = 'ctrl_cell_tf_idf'

RANDOM_MAPPING_CONTROL_TO_CONTROL = False

class HFDatasetForRNA(hf_dataset: Dataset, ctrl_name: str = 'control', gene_embs_file: str = './ESM2_pert_features.pt', mse_weights_pkl_file: str | None = None)

Bases: HFBaseDataset

MAIN_INPUT_NAME = 'ctrl_cell_expr'

RANDOM_MAPPING_CONTROL_TO_CONTROL = False