scanvi_explainer.utils.train_test_group_split

scanvi_explainer.utils.train_test_group_split#

scanvi_explainer.utils.train_test_group_split(adata: AnnData, groupby: str, train_size: float = 0.8, batch_size: int = 128, layer: str = 'counts') → tuple[dict[str, Tensor], dict[str, Tensor]]#

Function to split anndata object 80/20 per group in format required for SCANVIDeep explainer.

Bigger datasets might not fit to the GPU memory. To overcome this issue we recommend setting batch_size to 128, meaning each group will only use randomly sampled 128 cells. This speeds up the explainer at the cost of correctness. We suggest bootstrapping this process multiple times.

Parameters:

  • adata – AnnData Annotated dataset

  • groupby (str) – Column in metadata by which the dataset should be split by

  • train_size – float Training size (background), by default 0.8

  • batch_size – int Number of cells used from each group, by default 128

Returns:

Train and test splits

Return type:

tuple[dict[str, torch.Tensor], dict[str, torch.Tensor]]

Contents