KEGNI: knowledge graph enhanced framework for gene regulatory network inference.

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) are crucial for understanding cellular mechanisms.
• Inferring cell type-specific GRNs is essential but challenging with complex biological data.
• Existing methods often lack the ability to integrate prior biological knowledge effectively.

Purpose of the Study:

• To present KEGNI (Knowledge graph-Enhanced Gene regulatory Network Inference), a novel framework for GRN inference.
• To improve the accuracy and interpretability of GRN inference using single-cell RNA sequencing (scRNA-seq) data.
• To demonstrate KEGNI's capability in identifying driver genes and elucidating regulatory mechanisms.

Main Methods:

• KEGNI employs a knowledge-guided approach using a graph autoencoder.
• It integrates a knowledge graph to capture and infer gene regulatory relationships.
• The framework is designed to process scRNA-seq data, with potential for paired scRNA-seq and scATAC-seq data.

Main Results:

• KEGNI demonstrated superior performance over existing methods in GRN inference.
• The framework successfully identified driver genes across different cellular contexts.
• KEGNI effectively elucidated complex regulatory mechanisms using integrated knowledge.

Conclusions:

• KEGNI provides a powerful and accurate method for inferring cell type-specific GRNs.
• The knowledge-guided approach enhances the understanding of gene regulation from scRNA-seq data.
• KEGNI's modular design allows for flexible integration of diverse knowledge sources for various applications.