Probabilistic modelling of single-cell bisulfite sequencing data with MethylVI

2026-04-28 · Source: Nature Machine Intelligence · Field: Science & Research — Life Sciences & Biology, Mathematics & Computational Sciences · Depth: Expert, long

Summary

Methylation variational inference (MethylVI) is a new probabilistic modeling framework designed to analyze single-cell bisulfite sequencing (scBS-seq) data. This framework addresses the challenge of distinguishing genuine epigenetic variability from technical noise, such as sparse cytosine coverage, which often confounds scBS-seq datasets. MethylVI offers out-of-the-box capabilities for core analysis tasks, including dimensionality reduction, integration of data from various scBS-seq protocols, and identification of differentially methylated genes. The model integrates with existing single-cell analysis tools, demonstrated through extensions for single-cell reference atlas mapping and multiomic modeling. The code for MethylVI is available as part of the scvi-tools Python package, and reproducibility code is hosted on GitHub and Zenodo (ref. 94, 2026).

Key takeaway

For AI Scientists and Research Scientists working with single-cell epigenomic data, adopting MethylVI can significantly improve the accuracy of your analyses by explicitly accounting for technical noise inherent in scBS-seq. You should explore its integration capabilities with existing single-cell tools like scvi-tools to streamline workflows for tasks such as data integration and differential methylation analysis, potentially revealing clearer biological insights from complex datasets.

Key insights

MethylVI disentangles biological signals from technical noise in single-cell bisulfite sequencing data.

Principles

• Probabilistic modeling improves signal-to-noise in scBS-seq.
• Flexible frameworks enhance multi-task single-cell analysis.

Method

MethylVI employs a probabilistic modeling framework based on deep generative modeling techniques to explicitly separate biological signals from technical confounding factors in scBS-seq data.

In practice

• Perform dimensionality reduction on scBS-seq data.
• Integrate diverse scBS-seq protocol data.
• Identify differentially methylated genes.

Topics

• Single-cell Bisulfite Sequencing
• Methylation Variational Inference
• Probabilistic Modeling Framework
• Multiomic Data Integration
• Differentially Methylated Genes

Code references

• suinleelab/methylVI-reproducibility
• scverse/scvi-tools

Best for: AI Scientist, Research Scientist

Related on AIssential

• Dissecting epigenetic heterogeneity in single-cell DNA methylomes with a unified framework — scMethCraft accurately models single-cell DNA methylation data by addressing its unique characteristics with a hybrid neural network.
• Toward informed batch correction for single-cell transcriptome integration — Batch effects in single-cell data hinder comparisons; future methods need informed, interpretable modeling.
• scHelix: Asymmetric Dual-Stream Integration via Explicit Gene-Level Disentanglement — scHelix integrates scRNA-seq data by disentangling genes into invariant Anchors and sensitive Variants to preserve biological fidelity.
• scGTN: Deep Siamese Graph Transformer Network for Single-cell RNA Sequencing Clustering — scGTN integrates gene expression and intercellular structure via a Siamese graph transformer for robust scRNA-seq clustering.
• Conditional Monge Gap enables generalizable single-cell perturbation modelling — CMonge is a conditional optimal transport framework that accurately predicts single-cell perturbation responses and generalizes to unseen conditions.
• scCBGM: Interpretable Single-Cell Counterfactual Editing — scCBGM offers interpretable, precise counterfactual editing of single cells, addressing the combinatorial complexity of cellular phenotype mapping.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Nature Machine Intelligence.