Inferring cell-specific gene regulatory networks based on causal graph embedding.

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Single-cell resolution of gene regulatory networks (GRNs) is crucial for understanding cellular functions.
• Existing methods often neglect intercellular heterogeneity and global expression organization.
• Inferring accurate GRNs at the single-cell level remains a challenge.

Purpose of the Study:

• To develop a computational framework, CSGRN, for inferring gene regulatory networks (GRNs) at single-cell resolution.
• To improve the accuracy and robustness of GRN inference by integrating causal structures and cell-specific networks.
• To provide tools for analyzing regulatory dynamics and identifying key regulators in complex biological systems.

Main Methods:

• CSGRN is a supervised computational framework integrating graph embedding and conditional cell-specific networks (CCSNs).
• It infers GRNs for individual cells from single-cell RNA sequencing (scRNA-seq) data.
• Downstream analyses include signal flow analysis and gene perturbation simulation.

Main Results:

• CSGRN demonstrated superior performance compared to nine existing methods across three benchmark datasets.
• The framework successfully infers accurate and robust GRNs at single-cell resolution.
• Signal flow analysis and gene perturbation simulations revealed cell type-specific regulatory programs and key regulators.

Conclusions:

• CSGRN provides a powerful framework for inferring and analyzing gene regulatory networks from scRNA-seq data.
• The method effectively addresses intercellular heterogeneity and global expression organization.
• CSGRN facilitates the investigation of gene regulation in cellular differentiation and disease processes.