Elizabeth M. Perez

ENCODE-rE2G, a classifier trained on K562 CRISPR perturbations, maps 13,455,443 enhancer-gene interactions across 352 biosamples, matching or exceeding prior models on CRISPR, eQTL, and GWAS benchmarks.

The paper introduces CRISPRi-FlowFISH to perturb enhancers and an Activity-by-Contact model to predict enhancer-gene regulatory connections.

This study shows that cohesin loss eliminates loop domains, causing superenhancer links and gene downregulation, which are reversible upon cohesin restoration.