scvi.external.methylvi.METHYLVAE

class scvi.external.methylvi.METHYLVAE(n_input, contexts, num_features_per_context, n_batch=0, n_cats_per_cov=None, n_hidden=128, n_latent=10, n_layers=1, dropout_rate=0.1, likelihood='betabinomial', dispersion='region')

Bases: BaseModuleClass

PyTorch module for methylVI.

Parameters:

• n_input (int) – Total number of input genomic regions

• contexts (Iterable[str]) – List of methylation contexts (e.g. [“mCG”, “mCH”])

• num_features_per_context (Iterable[int]) – Number of features corresponding to each context

• n_batch (int (default: 0)) – Number of batches, if 0, no batch correction is performed

• n_cats_per_cov (Iterable[int] | None (default: None)) – Number of categories for each extra categorical covariate

• n_hidden (int (default: 128)) – Number of nodes per hidden layer

• n_latent (int (default: 10)) – Dimensionality of the latent space

• n_layers (int (default: 1)) – Number of hidden layers used for encoder and decoder NNs

• dropout_rate (float (default: 0.1)) – Dropout rate for neural networks

• likelihood (Literal['betabinomial', 'binomial'] (default: 'betabinomial')) – One of * 'betabinomial' - BetaBinomial distribution * 'binomial' - Binomial distribution

• dispersion (Literal['region', 'region-cell'] (default: 'region')) – One of the following * 'region' - dispersion parameter of BetaBinomial is constant per region across cells * 'region-cell' - dispersion can differ for every region in every cell

Attributes table

training

Methods table

generative(z, batch_index[, cat_covs])	Runs the generative model.
inference(mc, cov, batch_index[, cat_covs, ...])	High level inference method.
loss(tensors, inference_outputs, ...[, ...])	Loss function.
sample(tensors[, n_samples])	Generate observation samples from the posterior predictive distribution.

Attributes

METHYLVAE.training: bool

Methods

METHYLVAE.generative(z, batch_index, cat_covs=None)

Runs the generative model.

METHYLVAE.inference(mc, cov, batch_index, cat_covs=None, n_samples=1)

High level inference method.

Runs the inference (encoder) model.

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

Loss function.

METHYLVAE.sample(tensors, n_samples=1)

Generate observation samples from the posterior predictive distribution.

The posterior predictive distribution is written as \(p(\hat{x} \mid x)\).

Parameters:

• tensors – Tensors dict

• n_samples (default: 1) – Number of required samples for each cell

Return type:

dict[Tensor]

Returns:

x_new tensor with shape (n_cells, n_regions, n_samples)