Resolving how genetic perturbations act through gene regulatory paths

REGA-perturb resolves how CRISPR interference perturbations use TF–regulatory element–gene paths, inferring TF-target relationships with over fourfold gain versus random across 2.4 million cells.

• Why it matters: Expression-only Perturb-seq mixes contributions from many regulatory routes, so the specific regulatory path a perturbation follows remains unclear even with large models.
• What they did: REGA-perturb builds a path scaffold from reference epigenomes, then uses shared latent programs to fit perturbation RNA responses, inferring which TF–element–gene paths are used.
• The result: In three pooled CRISPRi screens (2.4 million cells), inferred TF-targets matched ChIP-seq binding, separated direct from indirect targets, predicted TF targets withheld from training, and captured TF activity shifts and perturbation-specific regulatory rewiring.

AI summary of the abstract · may contain errors · not medical advice · read the paper.