omicverse.space.Deconvolution

omicverse.space.Deconvolution#

class omicverse.space.Deconvolution(adata_sp, adata_sc=None)[source]#

Spatial deconvolution pipeline that aligns scRNA-seq references with spatial transcriptomics.

Parameters:

  • adata_sp (AnnData) – Spatial transcriptomics AnnData object (spots x genes).

  • adata_sc (AnnData or None) – Single-cell reference AnnData (cells x genes) containing cell-type labels.

Returns:

Initializes the deconvolution manager and backend placeholders.

Return type:

None

Examples

>>> decov = ov.space.Deconvolution(adata_sp=adata_sp, adata_sc=adata_sc)
>>> decov.deconvolution(method='Tangram', celltype_key_sc='cell_type')
__init__(adata_sp, adata_sc=None)[source]#

Initialize Deconvolution object. :type adata_sp: :param adata_sp: Spatial transcriptomics data. :type adata_sp: AnnData :type adata_sc: default: None :param adata_sc: Single-cell RNA-seq reference. If None, only spatial object is initialized. :type adata_sc: AnnData or None

Methods

__init__(adata_sp[, adata_sc])

Initialize Deconvolution object.

cell2location_inference([sample_kwargs])

Export cell2location posterior and derive normalized cell-type proportions.

deconvolution([method, celltype_key_sc, ...])

Infer spot-level cell-type composition from single-cell references.

impute([method])

Generate spot-level imputation outputs from a fitted spatial deconvolution model.

load_cell2location_model(mod_sp_path)

Load a previously trained cell2location spatial model.

load_tangram_model(model_path)

Load a pre-trained Tangram model for downstream projection/inference.

preprocess_sc([mode, n_HVGs, target_sum])

Preprocess the scRNA-seq reference before spatial mapping.

preprocess_sp([mode, n_svgs, target_sum, ...])

Preprocess spatial transcriptomics data and select spatially variable genes.

tangram_inference([sample_kwargs])

Infer spot-level cell-type proportions using a loaded Tangram model.
Contents