TDAGENE: Inference of Gene Regulatory Network Based on Topological Data Analysis and Graph Attention Network for...

Area of Science:

• Computational Biology
• Genomics
• Bioinformatics

Background:

• Single-cell RNA sequencing (scRNA-seq) enables high-resolution gene expression analysis.
• Inferring gene regulatory networks (GRNs) from scRNA-seq data is crucial for understanding cellular mechanisms.
• Existing GRN inference methods face challenges in capturing complex gene interactions.

Purpose of the Study:

• To propose a novel method, Topological Data Analysis-guided Gene Network Embedding (TDAGENE), for enhanced GRN inference.
• To integrate global topological features with local graph representations for improved gene expression modeling.
• To identify gene interaction relationships more effectively.

Main Methods:

• Developed TDAGENE, a method combining Topological Data Analysis (TDA) with graph embedding techniques.
• Integrated global topological features and local graph representations to capture GRN structure.
• Evaluated TDAGENE performance against existing GRN inference methods using benchmark datasets.

Main Results:

• TDAGENE demonstrated superior performance in GRN inference tasks.
• Achieved optimal predictions in 90% of datasets for Area Under the Precision-Recall Curve (AUPRC).
• Showed an average improvement of 17.66% in AUPRC and 3.08% in AUROC compared to state-of-the-art methods.
• Successfully applied TDAGENE to analyze key regulators (NANOG, SOX2, POU5F1) in cell fate specification.

Conclusions:

• TDAGENE effectively captures topological structures in GRNs, improving gene interaction identification.
• The method offers significant performance gains over existing GRN inference approaches.
• Incorporating topological information provides critical insights into cellular processes like cell fate specification.