Boosting single-cell gene regulatory network reconstruction via bulk-cell transcriptomic data.

Area of Science:

• Computational biology
• Genomics
• Bioinformatics

Background:

• Gene regulatory network (GRN) inference has shifted from bulk-cell to single-cell data analysis.
• Leveraging bulk-cell data for single-cell GRN prediction remains an underexplored area.

Purpose of the Study:

• To investigate the utility of bulk-cell data in enhancing single-cell GRN predictions.
• To develop a computational framework for integrating bulk and single-cell data for GRN inference.

Main Methods:

• A weakly supervised learning framework utilizing an axial transformer architecture was developed.
• Cell-type-specific GRNs were inferred using both single-cell RNA sequencing data and bulk-derived GRNs.

Main Results:

• Extensive experiments confirmed that bulk-cell transcriptomic data significantly improve single-cell GRN prediction accuracy.
• The proposed GRN-transformer method outperformed existing supervised and unsupervised approaches, achieving state-of-the-art results.
• The method successfully identified critical transcription factors and regulatory relationships associated with Alzheimer's disease risk genes.

Conclusions:

• Bulk-cell data is a valuable resource for improving single-cell GRN inference.
• The GRN-transformer offers a powerful and accurate tool for computational GRN recovery.
• This approach has implications for understanding complex diseases like Alzheimer's.