sfaira.train.SummarizeGridsearchEmbedding.get_gradients_by_celltype

SummarizeGridsearchEmbedding.get_gradients_by_celltype(model_organ: str, data_organ: str, organism: Optional[str], genome: Union[str, None, dict], model_type: Union[str, List[str]], metric_select: str, data_source: str, datapath, gene_type: str = 'protein_coding', configpath: Optional[str] = None, store_format: Optional[str] = None, test_data=True, partition_select: str = 'val', ignore_cache=False, min_cells=10)

Compute gradients across latent units with respect to input features for each cell type.

Parameters

• model_organ –

• data_organ –

• organism –

• model_type –

• metric_select –

• datapath –

• test_data –

• partition_select –

• ignore_cache –

• min_cells –

Returns

(cell types, input features) cumulative gradients