scvi.external.stereoscope.RNADeconv

class scvi.external.stereoscope.RNADeconv(n_genes, n_labels, **model_kwargs)

Bases: BaseModuleClass

Model of scRNA-seq for deconvolution of spatial transriptomics.

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

Parameters:

• n_genes (int) – Number of input genes

• n_labels (int) – Number of input cell types

• **model_kwargs – Additional kwargs

Attributes table

training

Methods table

generative(x, y)	Simply build the negative binomial parameters for every cell in the minibatch.
get_params()	Returns the parameters for feeding into the spatial data.
inference()	Inference.
loss(tensors, inference_outputs, ...[, ...])	Loss computation.
sample(tensors[, n_samples, library_size])	Sample from the model.

Attributes

RNADeconv.training: bool

Methods

RNADeconv.generative(x, y)

Simply build the negative binomial parameters for every cell in the minibatch.

RNADeconv.get_params()

Returns the parameters for feeding into the spatial data.

Return type:

tuple[ndarray]

Returns:

type list of tensor

RNADeconv.inference()

Inference.

RNADeconv.loss(tensors, inference_outputs, generative_outputs, kl_weight=1.0)

Loss computation.

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

Sample from the model.