Skip to main content
Fig. 1 | Genome Biology

Fig. 1

From: scVAEDer: integrating deep diffusion models and variational autoencoders for single-cell transcriptomics analysis

Fig. 1

scVAEDer overview. scVAEDer integrates VAE and DDM. First, a VAE is trained using the gene expression data. Then the VAE latent embedding is used to train the DDM through the processes of latent space diffusion and denoising. Combining together the model is able to decode back the gene space with high accuracy. scVAEDer can be used for different downstream analysis tasks such as generating novel high-quality scRNA-seq data, understanding changes in gene expression during cellular differentiation, predicting the effect of perturbations on new cell types when expression data is available for multiple conditions, detecting master regulators by interpolating between different cellular states and ranking fast responder genes based on their velocity values

Back to article page

Genome Biology

ISSN: 1474-760X

Contact us