GEEES: inferring cell-specific gene-enhancer interactions from multi-modal single-cell data.

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Gene-enhancer interactions are crucial for transcriptional regulation.
• Multi-modal single-cell datasets offer new avenues for studying these interactions at a cellular level.
• Existing computational methods have not been rigorously evaluated against benchmark datasets and often lack single-cell resolution.

Purpose of the Study:

• To develop and evaluate a novel computational approach, GEEES, for inferring gene-enhancer associations at the single-cell level.
• To comprehensively assess current methods for gene-enhancer interaction inference using multi-modal single-cell data.
• To identify key factors, such as gene-enhancer distance, that influence the accuracy of these inferences.

Main Methods:

• Developed GEEES (Gene EnhancEr IntEractions from Multi-modal Single Cell Data) for single-cell gene-enhancer interaction inference.
• Utilized multi-modal single-cell data (transcriptome and chromatin accessibility).
• Evaluated GEEES and alternative methods against benchmark datasets, incorporating gene-enhancer distance as a feature.

Main Results:

• GEEES infers gene-enhancer associations at the single-cell level.
• Incorporating gene-enhancer distance significantly improved the performance of all tested methods.
• Discrepancies were observed between inferred and gold-standard interactions, indicating limitations in current datasets and methods.
• GEEES offers enhanced cell representation learning for downstream analyses.

Conclusions:

• GEEES provides a valuable tool for single-cell gene-enhancer interaction inference.
• Gene-enhancer distance is a critical factor for improving accuracy.
• There is a pressing need for new high-throughput experiments to generate robust benchmark datasets for validating inferred gene-enhancer interactions.