HALO: hierarchical causal modeling for single cell multi-omics data.

Area of Science:

• Genomics
• Systems Biology
• Computational Biology

Background:

• Open chromatin often correlates with active transcription, but gene expression changes may not directly track chromatin accessibility shifts.
• Current single-cell multi-omics methods often focus on shared information, neglecting modality-specific dynamics and causal relationships.

Purpose of the Study:

• To propose HALO, a novel framework for modeling temporal causal relationships between epigenome and transcriptome.
• To differentiate between coupled (dependent) and decoupled (independent) changes in these modalities over time.

Main Methods:

• HALO employs a causal approach to model temporal relations at both representation and gene levels.
• It factorizes multi-omics data into coupled and decoupled latent representations.
• The framework matches gene-peak pairs and analyzes their temporal dynamics.

Main Results:

• HALO reveals the dynamic interplay between epigenome and transcriptome.
• It identifies analogous biological functions across modalities.
• The framework distinguishes lineage-specifying epigenetic factors and temporal cis-regulation interactions.

Conclusions:

• HALO provides a powerful approach to understand epigenome-transcriptome dynamics.
• It offers insights into cellular differentiation and disease mechanisms.
• The framework advances the analysis of single-cell multi-omics data by incorporating causal and temporal aspects.