Researchers have developed a new model called scE2G to map the complex regulatory connections between enhancers and genes in human cells. While millions of regulatory elements exist in the human genome, identifying which specific enhancers target which genes across different cell types has been a persistent challenge due to data limitations in complex tissues.
Previous methods relied on bulk measurements or unsupervised correlation techniques that often obscured precise regulatory patterns. By using single-cell ATAC-seq and multiomic data, scE2G provides a more accurate way to resolve these interactions. The model integrates multiple data points including chromatin accessibility, gene expression correlations, and distance metrics, trained on gold-standard CRISPR perturbation data.
In testing, scE2G outperformed existing single-cell models across multiple benchmarks. It remains stable across varying sequencing depths and cell counts, making it a reliable tool for researchers working with diverse experimental data. The team applied this model to map regulatory interactions in 45 distinct cell types, including immune cells, providing insights that bulk analysis previously missed.
This work is particularly useful for interpreting noncoding genetic variants associated with common diseases. By linking these variants to their target genes, scE2G offers a foundation for understanding how genomic variation influences health and disease at a cell-type-specific level. The researchers have made their model, benchmarking pipelines, and datasets available for the scientific community to advance gene regulation research.