Mapping enhancer-gene regulatory interactions from single-cell data.

Area of Science:

• Genomics
• Computational Biology
• Epigenetics

Background:

• Mapping enhancer-gene interactions is vital for understanding gene regulation and disease genetics.
• Accurate prediction of these interactions from single-cell data remains a significant challenge.

Purpose of the Study:

• To introduce scE2G, a new family of classification models for predicting enhancer-gene regulatory interactions.
• To leverage single-cell ATAC-seq and multiomic data for enhanced prediction accuracy.

Main Methods:

• Developed scE2G, a classification model trained on a large CRISPR perturbation dataset (>10,000 element-gene pairs).
• Utilized features from single-cell ATAC-seq and multiomic RNA and ATAC-seq data.
• Benchmarked scE2G against CRISPR perturbations, fine-mapped eQTLs, and GWAS variant-gene associations.

Main Results:

• scE2G demonstrated state-of-the-art performance in predicting enhancer-gene interactions across diverse cell types and perturbation categories.
• Applied scE2G to map regulatory interactions in heterogeneous tissues.
• Identified potential regulatory links between genes like INPP4B and IL15 and lymphocyte counts.

Conclusions:

• scE2G models provide a powerful tool for accurate enhancer-gene interaction mapping.
• This approach facilitates the interpretation of noncoding variants associated with complex human traits.
• The models enable the construction of regulatory maps across thousands of human cell types.