BayesCNet: Bayesian inference for cell type-specific regulatory networks leveraging cell type hierarchy in...

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Gene regulatory networks (GRNs) are crucial for understanding biological processes and diseases.
• Single-cell multiome technologies provide powerful tools for inferring cell type-specific GRNs by jointly profiling chromatin accessibility and gene expression.
• Current methods struggle with cellular heterogeneity and rare cell populations due to independent analysis or pseudo-bulk aggregation.

Purpose of the Study:

• To develop a novel computational method for robustly inferring cell type-specific GRNs from single-cell multiome data.
• To address the limitations of existing methods in resolving rare cell types and capturing cellular heterogeneity.
• To improve the accuracy of enhancer-gene linkage identification across diverse cell populations.

Main Methods:

• Introduction of BayesCNet, a Bayesian hierarchical model designed for joint inference of enhancer-gene linkages.
• Leveraging hierarchical relationships between cell types for enhanced information sharing.
• Validation through extensive simulations and application to real-world single-cell multiome datasets.

Main Results:

• BayesCNet consistently outperformed state-of-the-art methods in simulations, particularly for rare cell types.
• The method achieved higher accuracy in identifying enhancer-gene linkages, validated by promoter-capture Hi-C data.
• Reconstructed cell type-specific GRNs revealed key regulatory elements, demonstrating effective resolution of gene regulatory programs.

Conclusions:

• BayesCNet offers a powerful and accurate approach for reconstructing cell type-specific GRNs from single-cell multiome data.
• The model's ability to leverage hierarchical cell type information improves GRN inference, especially for challenging datasets with rare populations.
• This advancement facilitates a deeper understanding of gene regulation across cellular diversity and in disease contexts.