scvi-tools

To integrate scRNA-seq data from different batches, you can use deep generative models like scVI to remove technical variations and combine datasets. This process corrects batch effects to reveal true biological signals across multiple experiments.

Batch correction for single-cell RNA sequencing data works by applying deep generative models like scVI to remove technical variations. This process isolates true biological signals by conditioning on batch information provided in the AnnData object.

Before running scVI for single-cell data integration, you need to prepare AnnData objects containing raw counts and optional batch or label information. This structured data format is required to train the deep generative models effectively.

Yes, you can transfer cell type labels across multiple scRNA-seq datasets using the scANVI model. It leverages deep learning to accurately identify and annotate cell populations by transferring labels from a reference dataset.

The best way to identify distinct cell populations like T cells and B cells is using scANVI for cell type annotation. It leverages deep generative models to accurately transfer labels and identify specific cell types within integrated scRNA-seq data.

Your scVI model requires raw counts in the AnnData object because the deep generative model architecture is specifically designed to model unnormalized scRNA-seq data. Using normalized data would disrupt the underlying statistical assumptions needed for batch correction.