Inferring gene regulatory network from single-cell transcriptomes with graph autoencoder model.

Area of Science:

• Genomics
• Computational Biology
• Systems Biology

Background:

• Gene regulatory networks are crucial for cellular function, involving complex multi-gene interactions.
• Current single-cell gene regulatory network inference methods often overlook global structures, focusing only on pairwise gene relationships.
• Understanding global regulatory architecture is vital for deciphering complex biological systems.

Purpose of the Study:

• To develop a novel graph-based deep learning model for inferring gene regulatory networks from single-cell RNA-seq data.
• To address the limitations of existing methods by incorporating global regulatory structures.
• To accurately reconstruct gene regulatory networks and identify novel regulators.

Main Methods:

• Proposed DeepRIG (Deep learning model for Regulatory networks Inference among Genes), a graph-based deep learning approach.
• Constructed a prior regulatory graph by transforming gene expression data into a co-expression mode.
• Utilized a graph autoencoder to embed global regulatory information into gene latent embeddings for network reconstruction.

Main Results:

• DeepRIG accurately reconstructs gene regulatory networks.
• Demonstrated superior performance compared to existing methods on simulated and real biological networks.
• Successfully applied to human peripheral blood mononuclear cells and triple-negative breast cancer samples.

Conclusions:

• DeepRIG provides accurate cell-type-specific gene regulatory network inference.
• The model can identify novel regulators involved in disease progression and inhibition.
• This approach enhances the understanding of complex gene regulatory mechanisms in cellular systems.