Time-Varying Gene Regulatory Networks Inference Using KL Divergence from Single Cell Data

Area of Science:

• Computational Biology
• Genomics
• Systems Biology

Background:

• Reconstructing dynamic gene regulatory networks (GRNs) from time-series single-cell RNA sequencing (scRNA-seq) data is crucial for deciphering biological processes.
• Challenges include high dimensionality, data sparsity, and temporal heterogeneity inherent in scRNA-seq data.

Purpose of the Study:

• To develop a novel computational framework for accurate inference of time-varying GRNs from time-series scRNA-seq data.
• To address the limitations of existing methods in handling complex biological dynamics.

Main Methods:

• Integration of Kullback-Leibler (KL) divergence for temporal variation measurement with an autoregressive model.
• Application of various regularization techniques to infer gene interactions.
• Utilizing partial correlation analysis to determine the directionality (activation/inhibition) of regulatory relationships.

Main Results:

• The proposed framework successfully reconstructed dynamic network structures in both synthetic (10-gene) and experimental (THP-1 monocyte differentiation) datasets.
• Demonstrated accurate recovery of gene regulatory interactions and maintained temporal consistency.
• Validated the effectiveness of KL divergence and regularization in inferring time-varying networks.

Conclusions:

• The novel framework provides a robust method for inferring dynamic gene regulatory networks from time-series scRNA-seq data.
• This advancement facilitates a deeper understanding of biological system dynamics and gene regulation.
• The approach offers improved accuracy and temporal consistency compared to existing methods.