scvi.external.stereoscope.SpatialDeconv

class scvi.external.stereoscope.SpatialDeconv(n_spots, sc_params, prior_weight='n_obs')

Bases: BaseModuleClass

Model of single-cell RNA-sequencing data for deconvolution of spatial transcriptomics.

Reimplementation of the STModel module of Stereoscope [Andersson et al., 2020]: almaan/stereoscope.

Parameters:

• n_spots (int) – Number of input spots

• sc_params (tuple[ndarray]) – Tuple of ndarray of shapes [(n_genes, n_labels), (n_genes)] containing the dictionary and log dispersion parameters

• prior_weight (Literal['n_obs', 'minibatch'] (default: 'n_obs')) – Whether to sample the minibatch by the number of total observations or the minibatch size

Attributes table

training

Methods table

generative(x, ind_x)	Build the deconvolution model for every cell in the minibatch.
get_ct_specific_expression(y)	Returns cell-type-specific gene expression at the queried spots.
get_proportions([keep_noise])	Returns the loadings.
inference()	Inference.
loss(tensors, inference_outputs, ...[, ...])	Loss computation.
sample(tensors[, n_samples, library_size])	Sample from the model.

Attributes

SpatialDeconv.training: bool

Methods

SpatialDeconv.generative(x, ind_x)

Build the deconvolution model for every cell in the minibatch.

SpatialDeconv.get_ct_specific_expression(y)

Returns cell-type-specific gene expression at the queried spots.

Parameters:

y – cell types

SpatialDeconv.get_proportions(keep_noise=False)

Returns the loadings.

Return type:

ndarray

SpatialDeconv.inference()

Inference.

SpatialDeconv.loss(tensors, inference_outputs, generative_outputs, kl_weight=1.0, n_obs=1.0)

Loss computation.

SpatialDeconv.sample(tensors, n_samples=1, library_size=1)

Sample from the model.