In this episode, we hear from Romain Lopez and Gabriel Misrachi about

scVI—Single-cell Variational Inference.scVI is a probabilistic model for single-cell gene expression data thatcombines a hierarchical Bayesian model with deep neural networks encoding theconditional distributions. scVI scales to over one million cells and can beused for scRNA-seq normalization and batch effect removal, dimensionalityreduction, visualization, and differential expression. We alsodiscuss the recently implemented in scVI automatic hyperparameter selectionvia Bayesian optimization.

Links:

• Deep generative modeling for single-cell transcriptomics (Romain Lopez, Jeffrey Regier, Michael Cole, Michael I. Jordan, Nir Yosef)
• scVI on GitHub
• Should we zero-inflate scVI?
• Hyperparameter search for scVI
• Droplet scRNA-seq is not zero inflated (Valentine Svensson)