omicverse.bulk2single.Single2Spatial

omicverse.bulk2single.Single2Spatial(single_data: AnnData, spatial_data: AnnData, celltype_key: str, spot_key: list = ['xcoord', 'ycoord'], top_marker_num=500, marker_used=True, gpu: int | str = 0) → None[source]

Deep-learning mapper that projects single-cell profiles onto spatial coordinates.

Parameters:

  • single_data (anndata.AnnData) – Single-cell reference AnnData containing expression and cell types.

  • spatial_data (anndata.AnnData) – Spatial transcriptomics AnnData used as mapping target.

  • celltype_key (str) – Column name in single_data.obs containing cell-type labels.

  • spot_key (list) – Column names in spatial_data.obs storing x/y coordinates.

  • top_marker_num (int) – Number of marker genes used to train mapping model.

  • marker_used (bool) – Whether to restrict training features to marker genes.

  • gpu (Union[int,str]) – Compute device selector (CUDA index, 'mps', or CPU fallback).

Returns:

Initializes single-cell to spatial mapping workflow.

Return type:

None

Examples

>>> st_model = ov.bulk2single.Single2Spatial(single_data=single_data, spatial_data=st_data, celltype_key="Cell_type")