DeepGRNCS: deep learning-based framework for jointly inferring gene regulatory networks across cell subpopulations

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Gene regulatory networks (GRNs) are crucial for understanding cellular functions and diseases.
• Single-cell RNA sequencing (scRNA-seq) advances GRN inference accuracy.
• Existing methods often neglect intercellular heterogeneity and subpopulation similarities in scRNA-seq data.

Purpose of the Study:

• To develop a deep learning framework, DeepGRNCS, for jointly inferring GRNs across diverse cell subpopulations.
• To address limitations of current GRN inference methods that overlook cell heterogeneity.
• To improve the accuracy and scope of GRN inference from scRNA-seq data.

Main Methods:

• Proposed DeepGRNCS, a deep learning framework for joint GRN inference across cell subpopulations.
• Processed scRNA-seq data using the equal-width discretization method.
• Trained deep learning models to predict target gene expression from transcription factors (TFs) and inferred regulatory relationships.

Main Results:

• DeepGRNCS demonstrated superior performance compared to existing methods on simulated and real scRNA-seq datasets.
• The framework effectively predicts cell subpopulation-specific GRNs.
• Applied to non-small cell lung cancer data, identifying key genes and analyzing their biological relevance.

Conclusions:

• DeepGRNCS provides an effective approach for inferring gene regulatory networks specific to cell subpopulations.
• The method enhances the understanding of cellular mechanisms in complex biological systems.
• The source code is publicly available for broader research application.