DMDGRN: A data augmentation-based multilayer directed graph convolutional network for gene regulatory network...

Area of Science:

• Systems Biology
• Computational Biology
• Genomics

Background:

• Gene regulatory networks (GRNs) model interactions between transcription factors and target genes, crucial for cell identity and disease pathogenesis.
• Graph neural networks (GNNs) show promise for GRN inference but often overlook GRN characteristics like directionality and sparsity.
• Accurate GRN inference is a key challenge in understanding complex biological systems and disease mechanisms.

Purpose of the Study:

• To develop a novel GNN-based method for improved GRN inference.
• To address limitations of existing methods by incorporating directionality and sparsity considerations.
• To apply the developed method to identify therapeutic targets in breast cancer.

Main Methods:

• Propose DMDGRN, a data augmentation-based multilayer directed graph convolutional network for GRN inference.
• Utilize a phase matrix for the Laplacian operator to capture GRN directionality and track message propagation.
• Employ data augmentation and a multi-layer directed network with residual connections to handle sparsity and extract higher-order information.

Main Results:

• DMDGRN significantly improves GRN inference accuracy on benchmark datasets.
• The method successfully identifies relevant therapeutic candidates for human breast cancer.
• Evaluations confirm the effectiveness of data augmentation and multilayer directed architecture for GRN inference.

Conclusions:

• The adopted strategies in DMDGRN are effective for inferring gene regulatory networks.
• DMDGRN shows potential for uncovering disease-relevant regulatory mechanisms and identifying therapeutic targets.
• This method serves as a promising tool for advancing computational biology and translational medicine.