MuVi

MuVi is a multi-view latent variable model for the analysis of complex omics datasets. It enables the integration of multiple “views” (e.g. RNA, protein, metabolites) while incorporating domain-specific prior knowledge through predefined feature sets. MuVi disentangles biological variability from technical or confounding effects and provides interpretable latent factors. Originally developed as slalom for single-cell RNA-seq data, MuVi extends this concept to multi-omics data with structured sparsity and improved scalability.

Key Features

  • Multi-View Integration: Supports multiple data modalities simultaneously (e.g. RNA + protein) and models both shared and view-specific sources of variation.

  • Incorporation of Prior Knowledge: Allows the inclusion of gene sets, pathways, or other biological feature groups as structured priors that guide factor discovery.

  • Structured Sparsity: Uses hierarchical and feature-group-based sparsity priors to improve interpretability and efficiency in large-scale datasets.

  • Scalable and Efficient: The re-implementation in MuVi provides substantial performance improvements compared to slalom, enabling application to high-dimensional data.

  • Interpretation and Visualization Tools: Offers downstream functions such as explained-variance plots, latent-space clustering, and factor inspection utilities.

  • Flexible and Modern Implementation: Available as a Python package with compatibility for AnnData/MuData objects or standard DataFrames; GPU acceleration supported.

Intended Users

MuVi is designed for researchers in bioinformatics, systems biology, and multi-omics integration who work with large, heterogeneous datasets spanning several data modalities (e.g. transcriptomics, proteomics, metabolomics).

It is particularly useful if you aim to:

  • Identify latent biological factors linked to known pathways or gene sets while discovering new sources of variation,

  • Integrate multiple omics layers in a unified framework,

  • Obtain interpretable rather than purely black-box results, and

  • Work efficiently with large-scale single-cell or multi-omics data.

For simpler, single-view analyses (e.g. bulk RNA-seq only), other lightweight tools may be sufficient — but MuVi offers an ideal solution when biological interpretability and data integration are key.

Website
This email address is being protected from spambots. You need JavaScript enabled to view it.

Neither development nor maintenance funded by de.NBI