Single-cell multi-ome regression models identify functional and disease-associated enhancers and enable chromatin...

Area of Science:

• Genomics
• Computational Biology
• Molecular Biology

Background:

• Understanding gene regulation is crucial for deciphering cellular function and development.
• Existing methods for linking gene expression to chromatin accessibility have limitations.

Purpose of the Study:

• To develop a novel gene-level regulatory model for predicting single-cell gene expression.
• To link enhancers to their target genes using multi-omic sequencing data.
• To enable developmental trajectory analysis through chromatin potential fields.

Main Methods:

• Developed SCARlink, a gene-level regulatory model using regularized Poisson regression on tile-level accessibility data.
• Utilized multi-omic (scRNA-seq and scATAC-seq co-assay) sequencing data.
• Employed Shapley value analysis to identify cell-type-specific enhancers.

Main Results:

• SCARlink accurately imputes gene expression from chromatin accessibility, outperforming existing methods.
• Identified cell-type-specific enhancers validated by promoter capture Hi-C.
• Discovered significant enrichment of identified enhancers in fine-mapped eQTLs and GWAS variants.
• Demonstrated the utility of SCARlink-predicted gene expression for developmental trajectory analysis.

Conclusions:

• SCARlink provides a robust framework for gene regulation modeling using multi-omic data.
• The model effectively links enhancers to target genes and identifies functionally relevant regulatory elements.
• SCARlink facilitates advanced analyses such as developmental trajectory inference.