Model

Model#

The DRVI (Disentangled Representation Variational Inference) model is a model designed for single-cell omics data analysis. It provides disentangled latent representations that separate individual biological processes, enabling better interpretation and downstream analysis.

Overview#

DRVI extends the standard variational autoencoder architecture with specialized decoder architecture. The model learns disentangled representations and separates different sources of variation in the data, such as:

  • Biological factors: Cell types, developmental processes, perturbation responses, signaling pathways

  • Technical factors: Background expressions, technical stress responses

DRVI is now part of scvi-tools#

As of drvi-py version 0.3.0, the DRVI PyTorch model is no longer maintained in this package. It has been contributed to scvi-tools and lives there as scvi.external.DRVI (requires scvi-tools >= 1.5.0). New code should import the model directly:

from scvi.external import DRVI

For backward compatibility, drvi.model.DRVI remains importable as an alias for scvi.external.DRVI. See the scvi-tools DRVI documentation for the full model API.

Everything else in this package — the utility, plotting, metrics, and interpretability tools documented in the rest of this API reference — continues to be maintained here and works on top of the scvi-tools model.

model.DRVI

Disentangled Representation Variational Inference [].

Usage Example#

import anndata as ad
from scvi.external import DRVI

# Load your data
adata = ad.read_h5ad("your_data.h5ad")

# Setup anndata
DRVI.setup_anndata(
    adata,
    layer="counts",
    batch_key="batch",
)

# Initialize the model
model = DRVI(
    adata,
    n_latent=64,
    n_hidden=128,
    n_layers=2,
)

# Train the model
model.train(max_epochs=400)

# Get disentangled representations
latent = model.get_latent_representation()

# Please check tutorials for more details and downstream steps