scvi-basic

You can perform unsupervised integration and batch correction for scRNA-seq data using scVI to align datasets and remove batch effects while preserving biological variation. It constructs a shared latent space for cross-batch analyses.

Yes, scVI is applicable for preprocessing support before scANVI workflows. It trains on raw-count AnnData to obtain a latent representation, which serves as the foundation for downstream scANVI and differential expression analyses.

scVI supports flexible, GPU-accelerated training using scvi-tools, but GPU is not strictly required. It integrates with anndata and the scvi-framework to process raw counts and build latent representations efficiently.

Using deep learning frameworks like scVI is a highly effective way to remove batch effects in single-cell RNA-seq data. It applies unsupervised integration across batches to align datasets while preserving underlying biological variation.

Yes, you need raw counts in your AnnData object to build a latent space with scVI. The model requires raw count inputs to accurately train and generate batch-corrected expression and latent representations.

You should avoid unsupervised integration with scVI if your scRNA-seq data lacks raw counts or does not suffer from significant batch effects. It is specifically designed for cross-batch alignment and requires raw count inputs.