UniVI

UniVI is a Python toolkit for multimodal single-cell data integration using a mixture-of-experts β-variational autoencoder. It integrates heterogeneous measurements, including gene expression, chromatin accessibility and surface-protein abundance, into a shared latent representation. UniVI supports paired and partially observed multimodal datasets, cross-modal imputation, data denoising, synthetic data generation and reference-to-query mapping. It provides training and evaluation workflows, fitted preprocessing transformations and reproducible tutorials for single-cell multi-omics analysis.