bioRxivJClub
Regulatory grammar design of tissue-specific enhancers across species
bioRxiv · · Preprint · Open access
Yao, Tran + 7 more
Abstract ↗AI summary
The abstract is read at the publisher; the summary is JClub's.
REGAL generated tissue-specific enhancers across species by programming regulatory grammar, achieving predicted lineage-specific expression in transgenic mouse embryos and matching or exceeding leading designs at notably shorter length.
- Why it matters: Enhancer deep-learning approaches optimize sequences while leaving regulatory grammar implicit, limiting mechanistic design of tissue specificity and cross-species generalization.
- What they did: REGAL uses reinforcement learning to explicitly program enhancer grammar via sequence-to-function oracles, with a shared placement architecture and interchangeable oracles and motif vocabularies, then validates with experiments and perturbations.
- The result: Tissue specificity depends on coordinated target-tissue activation and off-target repression, requiring both motif composition and spatial organization, and REGAL-designed enhancers drive predicted lineage-specific expression in mouse embryos, outperform a nucleotide-level optimizer in human cells, and match or exceed leading designs in zebrafish and Drosophila at shorter length.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.
- Open access
Free. We’ll email you when a new paper cites this one or any other paper you follow on JClub — at most one email a day. Every email has a one-tap unsubscribe.