An encyclopedia of human enhancer-gene regulatory interactions

Area of Science:

• Genomics
• Gene Regulation
• Computational Biology

Background:

• Identifying enhancer-gene interactions is crucial for understanding gene regulation and the impact of genetic variations on diseases.
• Existing methods for mapping these interactions are limited in scale and accuracy.

Purpose of the Study:

• To create and evaluate a comprehensive resource of enhancer-gene regulatory interactions.
• To develop and validate a state-of-the-art predictive model for enhancer-gene mapping.
• To provide insights into the mechanisms of enhancer-promoter communication and its role in complex diseases.

Main Methods:

• Integrated predictive models, chromatin states, 3D genome contacts, and large-scale genetic perturbations from the ENCODE Consortium.
• Developed a systematic benchmarking pipeline using CRISPR perturbation data, fine-mapped expression quantitative trait loci (eQTLs), and genome-wide association study (GWAS) variants.
• Built and validated the ENCODE-rE2G predictive model using iterative perturbations and supervised machine learning.

Main Results:

• Generated a resource of over 92 million enhancer-gene regulatory interactions across 1,458 biosamples and 369 cell types.
• The ENCODE-rE2G model achieved state-of-the-art performance in predicting enhancer-gene relationships.
• Identified key features guiding enhancer-promoter communication, including enhancer activity, 3D contacts, promoter class, and enhancer-enhancer synergy.
• Improved the linking of noncoding variants to target genes and cell types for common complex diseases.

Conclusions:

• The developed resource and ENCODE-rE2G model offer a valuable tool for studying gene regulation and human genetics.
• The findings enhance our understanding of regulatory complexity across genes and the mechanisms underlying complex diseases.
• This work provides genome-wide maps and predictive models crucial for future research in gene regulation and disease association.