SCENT defines non-coding disease mechanisms using single-cell multi-omics

Genome-wide association studies (GWAS) have mapped hundreds of thousands of genomic locations that underlie disease susceptibility. However, only rarely is it possible to pinpoint causal variants or target genes from GWAS signals. More than 90% of loci act through non-coding, regulatory mechanisms, and predicting variant effects on gene expression has been extremely challenging. The translation of GWAS loci into disease mechanisms requires human tissue-specific maps of functionally active regulatory elements that harbour causal variants, as well as the genes they target. To this end, we developed SCENT (single-cell enhancer–target gene mapping), a tool to generate cell-type-specific enhancer–gene maps from single-cell multimodal ATAC-seq and RNA-seq data¹.

SCENT directly models associations between ATAC peak accessibility and RNA counts across individual single cells in a specific cell type, rather than normalizing or aggregating them, as previous methods have done. This approach resulted in more accurate enhancer–gene linkages, as validated by expression quantitative loci, CRISPR-Flow FISH and H3K27ac HiChIP. Moreover, SCENT had greater enrichment for known disease alleles from statistical fine-mapping of GWAS than the bulk-tissue enhancer–gene maps conventionally used to annotate GWAS loci.