Inferring gene regulatory networks with graph convolutional network based on causal feature reconstruction

Area of Science:

• Computational biology
• Bioinformatics
• Systems biology

Background:

• Inferring gene regulatory networks (GRNs) is essential for understanding cellular mechanisms.
• Existing deep learning methods face challenges like information loss during network inference.

Purpose of the Study:

• To develop a novel deep learning approach for accurate GRN inference.
• To integrate causal inference with Graph Convolutional Networks (GCNs) for improved GRN reconstruction.

Main Methods:

• Utilized a Graph Convolutional Network (GCN) guided by causal information.
• Employed transfer entropy and a reconstruction layer for causal feature reconstruction.
• Implemented a Gaussian-kernel Autoencoder for efficient feature extraction from gene expression data.

Main Results:

• The proposed method demonstrated superior performance on DREAM5 and mDC datasets.
• Achieved higher Area Under the Precision-Recall Curve (AUPRC) metrics compared to existing algorithms.
• Causal feature reconstruction led to more reasonable, accurate, and reliable inferred GRNs.

Conclusions:

• The novel GCN-based approach with causal inference effectively infers gene regulatory networks.
• This method mitigates information loss and improves the biological relevance of inferred networks.
• The findings offer a more reliable tool for computational biology and bioinformatics research.